In this paper, we present a learning approach to goal assignment and trajectory planning for unlabeled robots operating in 2D, obstacle-filled workspaces. More specifically, we tackle the unlabeled multi-robot motion planning problem with motion constraints as a multi-agent reinforcement learning problem with some sparse global reward. In contrast with previous works, which formulate an entirely new hand-crafted optimization cost or trajectory generation algorithm for a different robot dynamic model, our framework is a general approach that is applicable to arbitrary robot models. Further, by using the velocity obstacle, we devise a smooth projection that guarantees collision free trajectories for all robots with respect to their neighbors and obstacles. The efficacy of our algorithm is demonstrated through varied simulations. A video describing our method and results can be found here.

We address the problem of high speed autonomous navigation of quadrotor micro aerial vehicles with limited onboard sensing and computation. In particular, we propose a dual range planning horizon method to safely and quickly navigate quadrotors to specified goal locations in previously unknown and unstructured environments. In each planning epoch, a short-range planner uses a local map to generate a new trajectory. At the same time, a safe stopping policy is found. This allows the robot to come to an emergency halt when necessary. Our algorithm guarantees collision avoidance and demonstrates important advances in real-time planning. First, our novel short range planning method allows us to generate and re-plan trajectories that are dynamically feasible, comply with state and input constraints, and avoid obstacles in real-time. Further, previous planning algorithms abstract away the obstacle detection problem by assuming the instantaneous availability of geometric information about the environment. In contrast, our method addresses the challenge of using the raw sensor data to form a map and navigate in real-time. Finally, in addition to simulation examples, we provide physical experiments that demonstrate the entire algorithmic pipeline from obstacle detection to trajectory execution.

In this work, we consider the labeled multi-robot planning problem. In this paradigm, a team of robots at fixed start positions must navigate to pre-specified and noninterchangable goal positions. While many algorithms have been proposed for finding optimal solutions to this problem, most methods assume that the robots are kinematic agents, whereas in reality, robots often have high-order dynamics that must be respected by their trajectories. Here, we propose a centralized method for generating trajectories for teams of robots with general nth-order dynamics navigating to labeled goals. Our algorithm is safe and complete and additionally allows for decoupled optimization of each robot's trajectory as a Quadratic Program with linear constraints. We present simulation results for teams of up to 20 robots.

In this paper, we present a trajectory planning method to navigate a quadrotor with a cable-suspended payload through known obstacle-filled environments. We model the system as a hybrid dynamical system and formulate the trajectory generation problem as a Mixed Integer Quadratic Program (MIQP). Specifically, we address two novel challenges. First, we plan for a multi-body system, and obstacle avoidance must be guaranteed for the quadrotor, load, and the cable. Second, our method accommodates transitions between subsystems of the hybrid dynamical system, allowing for maneuvers that would otherwise be infeasible if the cable were constrained to remain taut. Numerical and experimental results validate the proposed approach for the full hybrid system.

This paper presents a system of multiple coordinating autonomous underwater vehicles (AUV) that can localize and track a shark tagged with an acoustic transmitter. Each AUV is equipped with a stereo-hydrophone system that provides measurements of the relative bearing to the transmitter, as well as an acoustic modem that allows for inter-AUV communication and hence cooperative shark state estimation and decentralized tracking control. Online state estimation of the shark's state is performed using a Particle Filter in which measurements are shared between AUVs. The decentralized control system enables the AUVs to circumnavigate a dynamic target, (i.e. the estimated shark location). Each AUV circles the target by tracking circles of different radii and at different phase angles with respect to the target so as to obtain simultaneous sensor vantage points and minimize chance of AUV collision. A series of experiments using two AUVs were conducted in Big Fisherman's Cove in Santa Catalina Island, CA and demonstrated the ability to track a tagged leopard shark (Triakis semifasciata). The performance of the tracking was compared to standard manual tracking performed using an directional hydrophone operated by a researcher in a boat. In an additional experiment, the AUVs tracked an acoustic tag attached to the tracking boat to quantify the error of the state estimation of the system.