Ravi N. Haksar

Ravi N. Haksar

Auto-generated profile, based on publicly available information

Ravi N. Haksar

Department of Mechanical Engineering, Stanford University, Stanford, CA, USA

Research Focus & Metrics

Academic Career Length

9Years (2018-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
1 October 2018
Verified
Conference Paper
Closed access
RHRavi N. Haksar
Ravi N. Haksar
MSMac Schwager
Mac Schwager
This paper proposes a distributed deep reinforcement learning (RL) based strategy for a team of Unmanned Aerial Vehicles (UAVs) to autonomously fight forest fires. We first model the forest fire as a Markov decision process (MDP) with a factored structure. We consider optimally controlling the forest fire without agents using dynamic programming, and show any exact solution and many approximate solutions are computationally intractable. Given the problem complexity, we consider a deep RL approach in which each agent learns a policy requiring only local information. We show with Monte Carlo simulations that the deep RL policy outperforms a hand-tuned heuristic, and scales well for various forest sizes and different numbers of UAVs as well as variations in model parameters. Experimental demonstrations with mobile robots fighting a simulated forest fire in the Robotarium at the Georgia Institute of Technology are also presented.

2 January 2023 in IEEE Transactions on Robotics
Verified
Journal Article
Closed access
OSOla Shorinwa
Ola Shorinwa
RHRavi N. Haksar
Ravi N. Haksar
PWPatrick Washington
Patrick Washington
MSMac Schwager
Mac Schwager
In this article, we present a distributed algorithm to solve a class of multirobot task assignment problems. We formulate task assignment as a mathematical optimization and solve for optimal solutions with a variant of the consensus alternating direction method of multipliers (C-ADMM). We provide C-ADMM-based algorithms for both the primal and dual problem formulations and show the advantages of each form depending on the problem specifics. In our algorithm, each robot solves a series of local optimization problems and communicates the results to its local neighbors, ultimately converging to an optimal task assignment in problems with linear objective functions and an optimal solution of the relaxed problem in convex problems. While many other distributed algorithms require a central station for their implementation, in our algorithm, each robot only communicates with its one-hop neighbors. In linear task assignment problems, our algorithm converges to the optimal task assignment, unlike many other distributed algorithms for this problem, which yield suboptimal solutions. We demonstrate our algorithms in task assignment problems over a variety of communication network topologies, where we show that our inexact dual algorithm is at least 60% faster than other distributed algorithms, which produce an optimal task assignment. In addition, our dual algorithm attains a 69% speedup in computation time compared to a notable distributed variant of the Hungarian method (Chopra et al., 2017). We also apply our algorithm to a multi-unmanned-aerial-vehicle persistent surveillance problem, showing its suitability for problems involving periodic task assignments.

18 February 2020 in IEEE Robotics and Automation Letters
Verified
Journal Article
Closed access
RHRavi N. Haksar
Ravi N. Haksar
STSebastian Trimpe
Sebastian Trimpe
MSMac Schwager
Mac Schwager
In this work, we develop a novel decentralized coordination algorithm for a team of autonomous unmanned aerial vehicles (UAVs) to surveil an aggressive forest wildfire. For dangerous environmental processes that occur over very large areas, like forest wildfires, multi-agent systems cannot rely on long-range communication networks. Therefore, our framework is formulated for very restrictive communication constraints: UAVs are only able to communicate when they are physically close to each other. To accommodate this constraint, the UAVs schedule a time and place to meet in the future to guarantee that they will be able to meet up again and share their belief of the wildfire state. In contrast with prior work, we allow for a discrete time, discrete space Markov model with a large state space as well as restrictive communication constraints. We demonstrate the effectiveness of our approach using simulations of a wildfire model that has $10^{298}$ total states.

17 February 2022
Verified
Book Chapter
Closed access
PDF file
RHRavi N. Haksar
Ravi N. Haksar
OSOlaoluwa Shorinwa
Olaoluwa Shorinwa
PWPatrick Washington
Patrick Washington
MSMac Schwager
Mac Schwager

1 December 2018
Verified
Conference Paper
Closed access
RHRavi N. Haksar
Ravi N. Haksar
MSMac Schwager
Mac Schwager
We consider controlling graph-based MDPs (GMDPs) with two special properties: (i) Anonymous Influence and (ii) Symmetry. Large-scale spatial processes such as wildfires, disease epidemics, opinion dynamics, and robot swarms are well modeled by GMDPs with these properties. We derive two efficient and scalable algorithms for computing approximately optimal control policies for large GMDPs with Anonymous Influence and Symmetry and derive sub-optimality bounds for these policies. Unlike prior work, our algorithms explicitly enforce a global control capacity constraint. Our methods scale linearly in the number of equivalence classes in the GMDP rather than the total number of MDPs in the graph. We demonstrate our methods in simulations of controlling a wildfire with a global fire retardant constraint and controlling an Ebola outbreak with a global medicine constraint. Our Ebola model is derived from data from the 2014 West Africa outbreak.

6 February 2023 in IEEE Transactions on Automatic Control
Verified
Journal Article
Closed access
RHRavi N. Haksar
Ravi N. Haksar
MSMac Schwager
Mac Schwager
We consider controlling a graph-based Markov decision process (GMDP) with a control capacity constraint given only uncertain measurements of the underlying state. We also consider two special structural properties of GMDPs, called Anonymous Influence and Symmetry. Large-scale spatial processes such as forest wildfires, disease epidemics, opinion dynamics, and robot swarms are well-modeled by GMDPs with these properties. We adopt a certainty-equivalence approach and derive efficient and scalable algorithms for estimating the GMDP state given uncertain measurements, and for computing approximately optimal control policies given a maximum-likelihood state estimate. We also derive sub-optimality bounds for our estimation and control algorithms. Unlike prior work, our methods scale to GMDPs with large state-spaces and explicitly enforce a control constraint. We demonstrate the effectiveness of our estimation and control approach in simulations of controlling a forest wildfire using a model with $10^{1192}$ total states.