Deep Learning for Robotic Mass Transport Cloaking

Deep Learning for Robotic Mass Transport Cloaking

Reza Khodayi-mehr and Michael M. Zavlanos Reza Khodayi-mehr and Michael M. Zavlanos are with the Department of Mechanical Engineering and Materials Science, Duke University, Durham, NC 27708, USA, {reza.khodayi.mehr, michael.zavlanos}@duke.edu.
Abstract

We consider the problem of Mass Transport Cloaking using mobile robots. The robots move along a predefined curve that encloses the safe zone and carry sources that collectively counteract a chemical agent released in the environment. The goal is to steer the mass flux around a desired region so that it remains unaffected by the external concentration. We formulate the problem of controlling the robot positions and release rates as a PDE-constrained optimization, where the propagation of the chemical is modeled by the Advection-Diffusion (AD) PDE. We use a Deep Neural Network (NN) to approximate the solution of the PDE. Particularly, we propose a novel loss function for the NN that utilizes the variational form of the AD-PDE and allows us to reformulate the planning problem as an unsupervised model-based learning problem. Our loss function is discretization-free and highly parallelizable. Unlike passive cloaking methods that use metamaterials to steer the mass flux, our method is the first to use mobile robots to actively control the concentration levels and create safe zones independent of environmental conditions. We demonstrate the performance of our method in simulations.

I Introduction

From natural disasters like wildfires [1] to environmental pollution [2] and chemical leaks [3, 4], many life-threatening processes can mathematically be modeled as Distributed Parameter Systems (DPSs) using transport equations like the Advection-Diffusion (AD) PDE [5]. Creating safe zones in any of these domains, where the temperature or concentration are maintained within the survival limits for humans or animals, is a problem of paramount significance. The goal is to ensure that these zones remain unaffected by external environmental conditions. In this paper, we propose to use mobile robots to create such safe zones, or cloaks, in mass transport systems modeled by time-dependent AD-PDEs.

Cloaking in transport systems is an active area of research. Examples include mass transport cloaking [6], thermal transport cloaking [7], acoustic cloaking [8], and cloaking for elasticity [9]. The proposed methods typically rely on metamaterials to passively alter the transport phenomenon around a desired region, decreasing in this way the effect of the external environment. Mathematically, this is achieved by exploiting the invariance of the associated PDEs under curvilinear transformations. However, these methods generally cannot maintain desired environmental conditions in the cloaked zone. This is the case in [6] that creates safe zones in mass transport systems by enclosing those zones in a cloak consisting of cocentric spheres of varying diffusivities. While the concentration inside the safe zone is maintained below the levels of the surrounding environment, it cannot be precisely controlled. This limitation can be alleviated using active cloaking methods, specifically using robots that carry sources that counteract the release of a chemical agent in the environment. By controlling the position of the robots and the release rates of their sources, safe zones of desired concentration levels can be created, even under changing external environmental conditions. To the best of our knowledge, this is the first work to consider the problem of active cloaking in mass transport systems using mobile robots.

We model the propagation of the chemical agent in the environment by a time-dependent Advection-Diffusion (AD) PDE and assume that the robots move on predefined curves that enclose the desired safe zone. Then, we formulate the proposed active cloaking problem as a PDE-constrained optimization problem whose solution returns collision-free optimal robot trajectories and corresponding source release rates that maintain desired concentration levels in the safe zone. To solve this optimization problem, we propose a new method that employs Deep Neural Networks (NNs) to approximate the solution of the PDE. A major contribution of the proposed method is a novel loss function used to train the NN that employs the variational form of the PDE as opposed to its differential form. Given this loss function, we formulate the robotic Mass Transport Cloaking (MTC) problem as an unsupervised model-based learning problem that we solve using state-of-the-art stochastic gradient descent algorithms [10]. Common model-based learning approaches in the robotics literature modify a known model using large amounts of labeled data collected before training, e.g., samples of robot trajectories, to account for uncertainties and unmodeled dynamics in the model; see e.g., [11]. To the contrary, here we train the NN to approximate the solution of the AD-PDE using unlabeled sample points of the space-time generated during training. The physics captured by the PDE guide the learning of the NN parameters.

The task of creating safe zones has been considered in the robotics literature, e.g., in containment control [12, 13] and perimeter patrol [14, 15, 16] problems. In the former, a network of robots are driven to a target destination while contained within a polygon created by the leaders. A possible application is to secure and remove hazardous materials without contaminating the surroundings [13]. In the latter, the objective is to prevent an adversary from penetrating an area of interest by periodically monitoring its boundaries using a team of mobile robots [15]. A closely related problem is considered in [17] where the objective is to contain a herd of animals within a safe zone. Unlike these problems, here we are interested in using mobile robots to create safe zones in DPSs where the time-dependent transport phenomenon is governed by PDEs.

The MTC problem is closely related to disturbance control of DPSs [18] and Source Identification (SI) [19, 20, 21]. The objective in the former is to maintain the equilibrium of a DPS against an exogenous disturbance moving in the domain whereas in the latter it is to identify a source function and a corresponding concentration field that matches an observed set of concentrations. Similar to SI, we formulate the MTC problem as a PDE-constrained optimization problem [22]. The difference is that in the MTC problem the unknown source terms are constrained by the predefined robot curves and the goal is to match the concentration at the safe zone to desired values. PDE-constrained optimization problems have been studied extensively in the literature [22]. A major challenge in solving these problems is dealing with the size of the discretized model obtained using numerical methods like the FE method [23]. This so-called curse of dimensionality is severely exacerbated when the problem is time-dependent. The only viable solution, currently available, is to use model reduction techniques but these approaches compromise accuracy and stability for speed [24].

In this paper, we propose an alternative approach to the MTC problem that employs Deep Neural Networks (NNs) to approximate the solution of the AD-PDE constraint. Using NNs we can incorporate the AD-PDE constraint in the loss function and obtain a penalty method that circumvents the computational challenges of discretization-based methods to the MTC problem. Specifically, these methods need to reconstruct and exactly enforce new time-dependent AD-PDE constraints at every iteration of the optimization process when the robot positions or release rates change. Unlike discretization-based methods, our learning-based approach is highly parallelizable and it can be easily extended to account for different objectives and constraints. An overview of different approaches for solving PDEs using NNs can be found in [25, Ch. 04]. One group of approaches utilize NNs to memorize the solution of PDEs. Particularly, they solve the PDE using a numerical method to obtain labeled training data and often utilize Convolutional NNs, as powerful image processing tools, to capture the numerical solution in a supervised learning way [26]. These approaches do not replace numerical methods but rather rely on them and introduce an extra layer of approximation. There also exist methods, called FE-NNs, which represent the governing equations at the element level using artifical neurons [27]. FE-NNs scale with the number of discretization points and are similar in spirit to numerical methods.

Most closely related to the method proposed in this paper are approaches that also directly train a NN to approximate the solution of the PDE in an unsupervised learning process. One of the early works of this kind is [28] that uses the residual of the PDE to define the required loss function. In order to remove the constraints from the training problem, the authors only consider simple domains for which the Boundary Conditions (BCs) can be manually enforced by a change of variables. Although these approaches attain comparable accuracy to numerical methods, they are impractical since, in general, enforcing the BCs might be as difficult as solving the original PDE. Following a different approach, the work in [29] utilizes a constrained back-propagation algorithm to enforce the initial and boundary conditions during training. In order to avoid solving a constrained training problem, the authors in [30] add the constraints corresponding to BCs to the objective as penalty terms. Similarly, in [31] the authors focus on the solution of PDEs with high dimensions using a Long Short-Term Memory architecture. Note that none of the above methods of solving PDEs using NNs, focus on controlling those PDEs nor do they involve mobile robots in any way.

Compared to the literature discussed above, the contributions of this work can be summarized as follows. To the best of our knowledge, this work is the first in the robotics literature to consider the active Mass Transport Cloaking (MTC) problem. Compared to passive cloaking methods that employ metamaterials to steer the transport phenomenon, our active MTC method can more precisely control the concentration levels in the safe zone. We propose a new approach to this problem that relies on Deep Learning to circumvent the computational challenges involved in using discretization-based methods to solve time-dependent PDE-constrained optimization problems. A major contribution of our proposed learning-based method is a novel loss function that we define to train the NN in an unsupervised model-based way, using the variational form of the PDE. The advantages of the proposed loss function, compared to existing approaches that use the residual of the PDE, are two-fold. First, it contains lower order derivatives so that the solution of the PDE can be estimated more accurately. Note that it becomes progressively more difficult to estimate a function from its higher order derivatives since differential operators are agnostic to translations. Second, it utilizes the integral (variational) form of the PDE, as opposed to its differential form, that considers segments of space-time as opposed to single points and imposes fewer smoothness requirements on the solution. Note that variational formulations have been successfully used in the FE method for a long time [23]. Compared to model-based learning approaches in the robotics literature that rely on supervised learning and large amounts of labeled data, here we propose an unsupervised method that relies on the physics captured by the PDE to guide the training. Finally, we validate the cloaking ability of the robot controllers returned by our method using a forward FE solution of the time-dependent AD-PDE. Note that although here we consider a specific control problem for the AD-PDE, the proposed principals generalize to other control problems subject to arbitrary dynamics.

The remainder of this paper is organized as follows. In Section II we discuss the AD-PDE and formulate the MTC problem. Section III-A is devoted to the solution of the time-dependent MTC problem using NNs. We present our results in Section IV and finally, Section V concludes the paper.

Ii Problem Formulation

Ii-a Advection-Diffusion PDE

Let denote a domain of interest where is its dimension and let denote a point in this domain and denote the time variable. Furthermore, consider a velocity vector field and its corresponding diffusivity field . Then, the transport of a quantity of interest , e.g., a chemical concentration, in this domain is described by the time-dependent Advection-Diffusion (AD) PDE [5]

(1)

where denotes the time derivative of the concentration and is the time-dependent source field that models any chemical reaction or mechanical action that leads to the release or collection of the chemical [32].

Given an appropriate set of Initial Conditions (ICs) and Boundary Conditions (BCs), it can be shown that the AD-PDE (1) is well-posed and has a unique solution [5]. In this paper, we use the following IC and BCs:

(2a)
(2b)

where for denote the boundary segments of and and possess appropriate regularity conditions [5]. Equation (2a) describes the state of the concentration field before the release of the chemical agent whereas equation (2b) prescribes the concentration value along the boundary segment as a time-varying function.

Ii-B Mass Transport Cloaking using Mobile Robots

Given the AD-PDE (1) and the domain of interest , consider a team of robots that carry sources that collectively counteract the release of the chemical. These mobile robots control the source term in the AD-PDE (1). The goal of the robots is to create a safe zone in the domain where the concentration is controlled to remain at a certain level. This can be achieved by formulating a planning problem that controls the robot positions in the domain as well as the release rates of their sources. Figure 1 shows a typical scenario where we assume the chemical agent is released outside and its effect is modeled by the BCs. 111Inclusion of the chemical release in the domain is equivalent to adding an extra source term in (1).

Fig. 1: A typical setup for the problem of Mass Transport Cloaking. The robots release or collect a chemical agent in an optimal rate along their paths , in response to changing environmental conditions, so that the concentration at the safe zone is maintained at a desired value.

In this figure, denotes the desired safe zone. The goal is to maintain the release of the chemical invisible to an observer inside . Specifically, let denote the number of robots. Then, the path of robot is denoted by for . We make the following assumptions: {assumption} [PDE Input Data] The geometry of the domain and its boundaries , the velocity and corresponding diffusivity fields, and the IC and BCs for are known.

The assumption that the domain is known is a reasonable one. Given the knowledge of the domain, the velocity and diffusivity fields can be estimated using Computational Fluid Dynamics or from measurements. Moreover, knowledge of the release of a chemical agent captured by the BCs, might be known for a specific task or can be measured.

{assumption}

[Robot Paths] The robot paths are constrained on a curve that is simple, differentiable, and does not intersect with the safe zone or other obstacles in the domain.

The curve often encloses and in many robotic applications corresponds to the perimeter of the safe zone [12, 14]. Given this curve, we can define the path of robot as

(3)

where and is a parameterization that maps the time interval to . Note that is not necessarily monotone, i.e., the robot can move back and forth along .

The release of the source, carried by robot , is limited to the location of the robot at any given time. In order to capture this local effect, we model the source term using a Gaussian function centered at the robot. Particularly, let denote the length-scale of the Gaussian function and the release rate of the source. Then, the source term in (1) is given by

(4)

where denotes the location of the robot at time .

{prob}

[Mass Transport Cloaking] Let denote the desired concentration level in the safe zone . Find the optimal paths and release rates to maintain concentration inside .

Problem 1 can be formulated as the following PDE-constrained optimization problem:

(5)

for , where in the relevant constraints, is the differential operator corresponding to the AD-PDE (1), and the objective of the optimization is defined as

(6)

Note that in equation (6), the spatial integration is performed only over the safe zone .

Iii Solution using Deep Neural Networks

In this section we first discuss the loss function that we use to train a Neural Network (NN) to approximate the solution of the time-dependent AD-PDE (1) for a given set of input data. Then, given this loss function, we address the Mass Transport Cloaking problem (5). The proposed loss function uses the variational formulation of the AD-PDE while our solution to problem (5) relies on adding this loss function, that captures the constraints in problem (5), to objective (6) in the form of a penalty term, to formulate an unconstrained training problem that can be solved using stochastic gradient descent methods.

Iii-a Neural Network Approximation of the AD-PDE

Let denote the weights and biases of the NN, a total of trainable parameters. Then the solution of the AD-PDE can be approximated by a nonlinear function , where maps the inputs and of the NN to its scalar output. The objective of training is to learn the parameters of the network so that approximates the solution of the AD-PDE (1) as well as possible. To capture this, we need to define a loss function that reflects how well the function approximates the solution of AD-PDE (1). A typical approach is to consider the residual of the PDE when is substituted into (1). This approach has two problems. First, for the AD-PDE (1) it requires evaluation of second order derivatives of . Estimating a function from its higher order derivatives is inefficient since differentiation only retains slope information and is agnostic to translation. Second, training the differential form of the PDE amounts to learning a complicated field by only considering the PDE-residual at training points, i.e., a measure-zero set, which is ineffective in capturing the physics of the transport phenomenon.

We follow a different approach here that addresses both of these issues. Particularly, we rely on the variational form of the PDE. Let be an arbitrary compactly supported test function. Multiplying (1) by and integrating over spatial and temporal coordinates we get

Performing integration by parts we have

Note that since is compactly supported over the temporal coordinate. Similarly,

where again the boundary terms vanish since is compactly supported. Finally, note that for velocities far below the speed of sound the incompressibility assumption holds and . Putting together all of these pieces, we get the variational form of the time-dependent AD-PDE as

(7)

This variational form only requires the first-order spatial derivative and also an integration over a non-zero measure set, i.e., the support of the test function, as opposed to a single point used in the strong form. The test function acts as a weight on the PDE residual and the idea is that if (7) holds for a reasonable number of test functions with their compact supports located at different regions in space-time , the function has to satisfy the PDE. A very important feature of the test function is that it is compactly supported. This allows local treatment of the PDE as opposed to considering the whole space-time at once and is the basis of the FE method [23].

Given the variational form (7), we can now define the desired loss function to approximate the solution of the AD-PDE (1) for a given set of input data. Consider a set of test functions sampling the space-time , a set of points corresponding to the IC, and sets of points for the enforcement of the BCs. Then, we define the loss function as

(8)

where is given by (7), stores the penalty weights corresponding to each term, and is the total number of training points for the BCs. Note that in the first term in (III-A), the integration is limited to the support of which is computationally very advantageous. Furthermore, this loss function is lower-bounded by zero. Since this bound is attainable for the exact solution of the AD-PDE (1), the value of the loss function is an indicator of how well the NN approximates the solution of the PDE.

Training using the loss function (III-A) is an instance of unsupervised learning since the solution is not learned from labeled data. Instead, the training data here are unlabeled samples of space-time and the physics captured by the AD-PDE (1) guides learning of the parameters . In that sense, our approach is a model-based method as opposed to a merely statistical method that automatically extracts features and cannot be easily interpreted. Moreover, unlike common robotic model-based learning approaches that modify a known model to account for the labeled data, obtained through numerous experiments, our approach proposes a novel use of NNs to learn the model from scratch without relying on labeled data.

{rem}

[Convergence] By the universal approximation theorem, for a large enough number of trainable parameters , the NN can approximate any smooth function [33]. Given that the AD-PDE (1) has a unique smooth solution given an appropriate set of input data, the NN should converge to this solution when ; see [31] for details.

{rem}

[Parallelization] Referring to (III-A), parallelization of the training process is trivial. We can choose the number of samples in accordance to the available computational resources and decompose and assign the summations to different processing units. The locations of the training points are also arbitrary and can be selected through random drawing or from a fixed grid over space-time.

Iii-B Mass Transport Cloaking using Deep Neural Networks

So far we have described how to approximate the solution of the AD-PDE (1) using a NN for a given set of input data. Next, we discuss how to use this approximation to solve the Mass Transport Cloaking (MTC) problem (5). The idea is to add the NN loss function that captures the constraints of the problem (5) to the objective in the form of a penalty term, and obtain an unconstrained optimization problem that can be solved using state-of-the-art stochastic gradient descent algorithms. Such algorithms have been very effective in the Deep Learning literature [10].

Specifically, to solve the PDE-constrained optimization problem (5), we parameterize the release rate by a polynomial where is the vector of coefficients corresponding to robot . We also parameterize in the definition of the robot paths (3) by the composition of another polynomial with parameters and the sigmoid function defined as Thus,

(9)

Furthermore, we use a set of points for the discrete approximation of the integral (6). Noting that the loss function (III-A) captures the constraints in (5), to solve the MTC problem we optimize the following objective function

(10)

Optimizing (10) provides the solution of the AD-PDE (1) for the time-dependent source term (4), with intensities specified by and paths specified by , that maintains the concentration at the desired level across the safe zone . Note that for simplicity we used the same length scale and number of parameters and for all robots in the source term (4); generalization is trivial.

{prop}

[Bounded Velocity] Robots following the paths (3) with parameterizations (9) have bounded velocities.

Proof.

Noting that , for the speed of the robot we have , where we drop the subscript for simplicity. Since is differentiable and defined over the compact set , it is bounded. Thus, we need to show that is bounded. To see this, note that

where is the polynomial that parameterizes the path of the robot. Then, since

and , we have Since is also continuously differentiable, it has to be bounded. Intuitively, when , the sigmoid function saturates and the robot approaches one end of the curve . ∎

Iv Numerical Experiments

In this section we present simulation results that illustrate our approach to solving the Mass Transport Cloaking (MTC) problem (5). The domain is depicted in Figure 1 where we select and . Before the release of the chemical agent, the system is at rest with zero concentration across the domain, i.e., in (2a). At time the release occurs causing the concentration to rise linearly at the corner of the boundaries as Note that the concentration reaches the maximum value of at and stays constant afterwards. For the boundary , the concentration decreases linearly to zero when moving from toward the corner between , i.e., in (2b) we have see Figure 2.

Fig. 2: BC value for as a function of spatial coordinate for two different time instances before and after .

A similar expression is used for along . The opposite boundaries are set to zero for all time, i.e., in (2b). Finally, we set the diffusivity field to a constant value of and define the velocity vector field for as

(11)

and , i.e., it has a constant magnitude but its direction rotates from to within and then stays constant. Note that variations in the BCs and velocity field stop at allowing to move toward steady-state.

We define the safe zone as a square centered at with side length of and set , i.e., we wish to maintain zero concentration at the safe zone as it was before the release of the chemical. In the following simulations, we use mobile robots that collaboratively cloak the safe zone . Referring to Assumption 1, we constrain the robots to move along a circular path with radius around . Particularly, the admissible path for robot is defined as

where is the parameterization corresponding to robot . In order to describe the source term (4), we use parameters for the release rates of the robots and parameters for their paths, where we fix the intercepts of the polynomials in (9) to specify the initial angular positions of the robots on the curve . These values correspond to where . We also set the characteristic length of the Gaussian source (4) to ; see Section III-B for details. Then, the robots optimally adjust their velocities along this curve and their release rates in response to the change in the BCs and the velocity field to maintain the concentration in . In order to prevent collision among the robots, we add the following penalty term (with an appropriate weight) to the training objective (10)

measuring the mutual angular distance of the robots, where determines the angle margin among robots and is set to in the following results.

We utilize the TensorFlow software to solve the desired MTC problem (5); see [34]. The test functions are selected to be trilinear FE basis functions for D hexagonal elements that are centered at arbitrary points in space-time; see Section III-A for more details and [23] for a description of FE basis functions. The numerical integration in (7) is performed using a two-point Gauss-Legendre quadrature at each dimension. We use a Multi-Layer Perceptron (MLP) NN architecture consisting of three dense hidden layers with neurons, respectively and sigmoid activation functions. In order to train the NN, we use samples of the temporal coordinate as well as samples of the spatial domain resulting in training points in the domain interior, training points for the IC, training points for each BC, and finally training points for the discrete approximation of the objective (6) at the safe zone . We set . Then, the unsupervised training for the MTC problem (5) is performed using the AdamOptimizer; see [10] for details. To validate the resulting optimal control inputs for the robots, we solve the AD-PDE (1) for these inputs using an in-house FE solver that we have developed using an explicit forward Euler scheme and DiffPack libraries [35, Ch. 3.10]. We use a mesh with nodes corresponding to the number of training points above.

Figure 3 shows the evolution of the concentration at obtained from the NN and FE solutions along with maximum and minimum (worst case) concentration values across the safe zone according to the FE solver.

Fig. 3: Concentration level at obtained from the NN and FE solutions as a function of time. The dashed lines delineate the maximum deviations from the desired value across the safe zone according to the FE solution.

It can be seen that the concentration stays close to the desired value . The highest concentration levels in are more than an order of magnitude smaller than the maximum concentration , meaning that external concentration levels are effectively invisible in the safe zone . Next, Figure 4 shows the optimal release rates and angular positions as functions of time.

(a) optimal release rates
(b) optimal angular positions
Fig. 4: Optimal source parameters as a function of time to maintain the concentration levels at at the specified level .

Note that the release rates in Figure 3(a) decrease to negative values in response to the growth of the concentration at the corner of the boundary . This corresponds to robots acting as sinks and collecting the extra chemical to maintain desired concentration levels in . Referring to Figure 3(b), the angular positions are adjusted in response to the rotation of the velocity field so that the safe zone is positioned between the robots. Not that since robot 2 is positioned downstream of the safe zone, for cloaking it relies on diffusive transport, which is as strong as advection here. Finally, Figure 5 shows snapshots of the concentration field, given by the NN approximation of the solution of the AD-PDE (1), at different time instances.

(a)
(b)
(c)
(d)
(e)
(f)
Fig. 5: Evolution of the concentration field as the solution of the AD-PDE (1) at different time instances. The black square denotes the safe zone whereas the black circle and star denote the location of robots , respectively. Moreover, the dashed circle shows the curve . Note that in all snapshots the level set passes through .

Note that the maximum concentration that occurs at corner , grows up to . After this time, the concentration field evolves toward steady-state and the variations become smaller. As a result, the designed optimal controls also reach a steady state after , as is evident from Figure 4. An animation of the optimal source field and the corresponding concentration field is given in [36]. Also an animation comparing the NN and FE solutions can be found in [37]. Note that the NN solution obtained by softly enforcing the PDE constraint in (5), closely matches the FE solution for the final optimal source field. Specifically, the L-error between the solutions is

Table I presents a comparative study of the performance of the NN as a function of the number of trainable parameters and the number of training points. Particularly, it summarizes the individual terms in the objective function (10). The first two cases correspond to a MLP with two hidden layers with neurons whereas the last NN, used in the previous simulations, has three hidden layers with neurons.

trainable parameters objective term concentration error variational form IC BCs
training points
objective value
training points
objective value
training points
objective value
TABLE I: Number of training points and averaged objective values per training point for the terms in objective (10).

The third column in this table reports the deviation of the predicted concentration from the desired value in the safe zone, given by objective (6). Comparing the first two cases, we observe that for a fixed architecture and number of trainable parameters , the normalized objective values get smaller as the number of training points increases. Moreover, comparing the last two cases, it can be seen that as the number of trainable parameters increases, the capacity of the NN increases, which amounts to smaller objective values. We can use larger values of to further minimize the training objective (10). Then, more training points should be used and techniques like regularization and dropout should be employed to prevent overfitting [10].

Note that based on the above simulations, a simple feed-forward MLP-NN with only trainable parameters continuously captures the D concentration field across space-time. Solving the same problem using the FE method would require a mesh with nodes and a separate interpolation for points outside the mesh. Note also that the proposed NN solution to the MTC problem can easily account for additional objectives and constraints, such as collision avoidance that is already incorporated in (10), temporally and spatially varying concentrations in the safe zone, minimization of control effort, e.g., traveled distance or release amount, parametric PDE input-data, and desired ranges (rather than exact values) of concentration in the safe zone. For instance, the integrand in objective (6) can be replaced with

where are weights penalizing concentrations above and below the desired value, respectively. If concentrations below are desirable, we set . Note that by setting , we recover the least-square objective (6). Finally, note that although our proposed learning-based method is very effective in solving the proposed MTC problem, it requires careful tuning of the penalty weights so that an acceptable balance between satisfaction of the PDE constraint and optimality of the desired robot controllers is obtained.

V Conclusion

In this paper we considered the problem of Mass Transport Cloaking (MTC) using teams of mobile robots that carry sources that counteract the release of a chemical in the environment. The goal of the robots is to steer the mass flux around a desired region, also called a safe zone, so that the zone remains unaffected by the external concentration. We utilized the AD-PDE to mathematically model the transport of a chemical in the domain and formulated the problem of planning the paths and release rates of the robots as a PDE-constrained optimization. We proposed a new approach to this problem using a Deep Neural Network to approximate the solution of the AD-PDE. Particularly, we defined a novel loss function based on the variational form of the PDE that facilitates the training process by lowering the differentiation order and using the integral form of the PDE as opposed to its differential form. Given this loss function, we reformulated the planning problem as an unsupervised model-based learning problem. We presented simulation results that demonstrate the ability of our method to solve the desired MTC problem.

References

  • [1] R. R. Murphy, R. Brown, R. Grant, and C. T. Arnett, “Preliminary domain theory for robot-assisted wildland firefighting,” in IEEE International Workshop on Safety, Security & Rescue Robotics, pp. 1–6, 2009.
  • [2] Oil in the sea III: inputs, fates, and effects. National Academies Press, 2003.
  • [3] A. S. Kalelkar, Investigation of large-magnitude incidents: Bhopal as a case study. Arthur D. Little, Incorporated, 1988.
  • [4] E. Lee, “Clinical manifestations of Sarin nerve gas exposure,” Journal of American Medical Association, vol. 290, no. 5, pp. 659–662, 2003.
  • [5] B. D. Reddy, Introductory functional analysis: with applications to boundary value problems and finite elements, vol. 27. Springer, 2013.
  • [6] S. Guenneau and T. Puvirajesinghe, “Fick’s second law transformed: one path to cloaking in mass diffusion,” Journal of The Royal Society Interface, vol. 10, no. 83, p. 20130106, 2013.
  • [7] X. Shen, Y. Li, C. Jiang, Y. Ni, and J. Huang, “Thermal cloak-concentrator,” Applied Physics Letters, vol. 109, no. 3, p. 031907, 2016.
  • [8] S. A. Cummer and D. Schurig, “One path to acoustic cloaking,” New Journal of Physics, vol. 9, no. 3, p. 45, 2007.
  • [9] G. W. Milton, M. Briane, and J. R. Willis, “On cloaking for elasticity and physical equations with a transformation invariant form,” New Journal of Physics, vol. 8, no. 10, p. 248, 2006.
  • [10] F. Chollet, Deep learning with Python. Manning Publications Co., 2017.
  • [11] F. Berkenkamp, R. Moriconi, A. P. Schoellig, and A. Krause, “Safe learning of regions of attraction for uncertain, nonlinear systems with gaussian processes,” in Proceedings of IEEE Conference on Decision and Control, pp. 4661–4666, IEEE, 2016.
  • [12] J. Mei, W. Ren, and G. Ma, “Distributed containment control for Lagrangian networks with parametric uncertainties under a directed graph,” Automatica, vol. 48, no. 4, pp. 653–659, 2012.
  • [13] M. Ji, G. Ferrari-Trecate, M. Egerstedt, and A. Buffa, “Containment control in mobile networks,” IEEE Transactions on Automatic Control, vol. 53, no. 8, pp. 1972–1975, 2008.
  • [14] Y. Elmaliach, N. Agmon, and G. A. Kaminka, “Multi-robot area patrol under frequency constraints,” Annals of Mathematics and Artificial Intelligence, vol. 57, no. 3-4, pp. 293–320, 2009.
  • [15] N. Agmon, S. Kraus, and G. A. Kaminka, “Multi-robot perimeter patrol in adversarial settings,” in Proceedings of IEEE International Conference on Robotics and Automation, pp. 2339–2345, IEEE, 2008.
  • [16] A. Marino, L. Parker, G. Antonelli, and F. Caccavale, “Behavioral control for multi-robot perimeter patrol: A finite state automata approach,” in Proceedings of IEEE International Conference on Robotics and Automation, pp. 831–836, IEEE, 2009.
  • [17] Z. Butler, P. Corke, R. Peterson, and D. Rus, “Virtual fences for controlling cows,” in Proceedings of IEEE International Conference on Robotics and Automation, vol. 5, pp. 4429–4436, IEEE, 2004.
  • [18] M. A. Demetriou, “Guidance of a moving collocated actuator/sensor for improved control of distributed parameter systems,” in Proceedings of IEEE Conference on Decision and Control, pp. 215–220, IEEE, 2008.
  • [19] R. Khodayi-mehr, W. Aquino, and M. M. Zavlanos, “Model-based sparse source identification,” in Proceedings of American Control Conference, pp. 1818–1823, July 2015.
  • [20] R. Khodayi-mehr, W. Aquino, and M. M. Zavlanos, “Model-based active source identification in complex environments,” IEEE Transactions on Robotics, 2019.
  • [21] R. Khodayi-mehr, W. Aquino, and M. M. Zavlanos, “Distributed reduced order source identification,” in Proceedings of American Control Conference, pp. 1084–1089, June 2018.
  • [22] M. Hinze, R. Pinnau, M. Ulbrich, and S. Ulbrich, Optimization with PDE constraints, vol. 23. Springer Science & Business Media, 2008.
  • [23] T. J. Hughes, The Finite Element method: linear static and dynamic finite element analysis. Courier Corporation, 2012.
  • [24] A. C. Antoulas, D. C. Sorensen, and S. Gugercin, “A survey of model reduction methods for large-scale systems,” Contemporary Mathematics, vol. 280, pp. 193–220, 2001.
  • [25] N. Yadav, A. Yadav, and M. Kumar, An introduction to Neural Network methods for differential equations. Springer, 2015.
  • [26] Y. Khoo, J. Lu, and L. Ying, “Solving parametric PDE problems with artificial Neural Networks,” arXiv preprint arXiv:1707.03351, 2017.
  • [27] P. Ramuhalli, L. Udpa, and S. S. Udpa, “Finite-Element Neural Networks for solving differential equations,” IEEE Transactions on Neural Networks, vol. 16, no. 6, pp. 1381–1392, 2005.
  • [28] I. E. Lagaris, A. Likas, and D. I. Fotiadis, “Artificial Neural Networks for solving ordinary and partial differential equations,” IEEE Transactions on Neural Networks, vol. 9, no. 5, pp. 987–1000, 1998.
  • [29] K. Rudd, Solving Partial Differential Equations Using Artificial Neural Networks. PhD thesis, Duke University, 2013.
  • [30] Y. Shirvany, M. Hayati, and R. Moradian, “Multilayer perceptron Neural Networks with novel unsupervised training method for numerical solution of the partial differential equations,” Applied Soft Computing, vol. 9, no. 1, pp. 20–29, 2009.
  • [31] J. Sirignano and K. Spiliopoulos, “DGM: A deep learning algorithm for solving partial differential equations,” arXiv preprint:1708.07469, 2017.
  • [32] C. Kuttler, “Reaction-diffusion equations with applications,” 2011.
  • [33] K. Hornik, M. Stinchcombe, and H. White, “Multilayer feedforward networks are universal approximators,” Neural Networks, vol. 2, no. 5, pp. 359–366, 1989.
  • [34] M. Abadi, P. Barham, J. Chen, Z. Chen, A. Davis, J. Dean, M. Devin, S. Ghemawat, G. Irving, M. Isard, et al., “Tensorflow: a system for large-scale machine learning.,”
  • [35] H. P. Langtangen, Computational partial differential equations: numerical methods and Diffpack programming, vol. 2. Springer Science & Business Media, 2013.
  • [36] “Animation of time-dependent optimal source field and corresponding concentration field for MTC problem.” https://vimeo.com/291029849.
  • [37] “Animation comparing the NN solution of the Advection-Diffusion PDE with the Finite Element solution.” https://vimeo.com/364115722.
Comments 0
Request Comment
You are adding the first comment!
How to quickly get a good reply:
  • Give credit where it’s due by listing out the positive aspects of a paper before getting into which changes should be made.
  • Be specific in your critique, and provide supporting evidence with appropriate references to substantiate general statements.
  • Your comment should inspire ideas to flow and help the author improves the paper.

The better we are at sharing our knowledge with each other, the faster we move forward.
""
The feedback must be of minimum 40 characters and the title a minimum of 5 characters
   
Add comment
Cancel
Loading ...
393302
This is a comment super asjknd jkasnjk adsnkj
Upvote
Downvote
""
The feedback must be of minumum 40 characters
The feedback must be of minumum 40 characters
Submit
Cancel

You are asking your first question!
How to quickly get a good answer:
  • Keep your question short and to the point
  • Check for grammar or spelling errors.
  • Phrase it like a question
Test
Test description