I. Generalized Velocity Obstacles (GVO)
Abstract:
We have developed a reactive, COLREGs-compliant planning algorithm for avoiding dynamic obstacles. The algorithm determines a control action that respects the unmanned surface vehicle (USV) dynamics, minimizes the probability of a collision, and optimizes the time needed to reach a local waypoint. The planner utilizes predictive motion models of obstacles when evaluating candidate control actions.
Published Work:
P. Švec, B. C. Shah, I. R. Bertaska, J. Alvarez, A. J. Sinisterra, K. von Ellenrieder, M. Dhanak, and S. K. Gupta. Dynamics-Aware Target Following for an Autonomous Surface Vehicle Operating under COLREGs in Civilian Traffic. IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS ’13), Tokyo, Japan, November 3-7, 2013. PDF
II. Adaptive Sampling-based Generalized Velocity Obstacles
Abstract:
We have developed a model-predictive, local trajectory planning algorithm for USVs operating in congested and highly dynamic traffic. The planner generalizes the Velocity Obstacle concept to systems with non-linear dynamics, nonholonomic constraints, and any form of low-level feedback controller. High planning performance is achieved by searching in a resolution-adaptive space of vehicle candidate motion goals for a motion goal that optimizes the arrival time to a given global goal. The algorithm gradually identifies, with increasing resolution and focused sampling, an approximate representation of spatiotemporal obstacle regions. The sampling of motion goals can be constrained to ensure the International Regulations for Prevention of Collisions at Sea (COLREGs). We have carried out experiments to demonstrate the practical feasibility of using the planner on a 14-foot (4.3 m) long catamaran USV operating in a harbor-like environment.
Published Work:
P. Švec, B.C. Shah, I.R. Bertaska, W. Klinger, A. J. Sinisterra, K. von Ellenrieder, M. Dhanak, and S.K. Gupta. Adaptive sampling based COLREGS-compliant obstacle avoidance for autonomous surface vehicles. ICRA 2014 Workshop on Persistent Autonomy for Marine Robotics (PAMR ’14), Hong Kong, China, June 2014. PDF
III. Dynamics-Aware Reactive Planning for UGV to Avoid Collisions with Dynamic Obstacles on Uneven Terrains
Abstract:
Collision avoidance is a key capability for autonomous ground vehicles and must respect the dynamic constraints of the vehicle. Avoiding collisions with dynamic obstacles requires a risk-aware planner and accurate estimates of current vehicle and terrain states. In addition, the planner must consider modifications to the intended path, generating alternatives in real-time. It must also consider regulating speed along the various modified paths, finding a trajectory that avoids collision while minimizing deviation from the intended trajectory. In this paper, we present such a real-time dynamics aware reactive trajectory generator which produces trajectories that avoid dynamic obstacle collision. Our planner, accounts for uncertainty in the obstacle position and velocities while evaluating alternative trajectories and uses conservative estimates of vehicle dynamic constraints to ensure collision risk is minimized. It also uses “deferred value binding”, to exploit more accurate estimates of the obstacle states as obstacles approach the vehicle. It automatically adjusts the number of options being evaluated based on the estimated time to collision. In order to ensure real-time performance, our planner handles multiple dynamic obstacles by either grouping them into a single composite obstacle in the configuration space or deals with them sequentially by prioritizing obstacles based on the estimated time to collision. We present simulation results to show that the planner is able to effectively deal with dynamic obstacles on terrains with varying slopes.
Brual Shah 