Navigation in Tight Environments

Navigation in Tight Environments

Maneuvering autonomous systems in an environment with obstacles is a challenging problem that arises in a number of practical applications, including robotic manipulators, self-driving cars and autonomous quadcopters. In all these applications, a fundamental feature is a system's ability to avoid collisions.

We have developed a framework that allows us to formulate navigation and collision-avoidance problems as an optimal control problem, where a cost can be minimized. In contrast to many existing approaches, our formulation is exact and results in a smooth optimization problem that (i) does not introduce conservatism, and (ii) can be solved using numerical solvers that employ gradient or interior-point algorithms.

The proposed framework was evaluated, in simulation, on a quadcopter navigation and automated parking problem, where the robots must navigate in tight environments. Our studies indicate that the proposed framework allows real-time optimization-based trajectory planning.

Contact:
 Xiaojing (George) ZHANG

Resources: Paper and Source Code

TrajQuad_3D_Video.gif
TrajReverseHOBCA25x.gif
TrajParallelHOBCA25x.gif
TrajTrailer_ParkingVideo.gif
G380: Our New Autonomous Research Vehicle

G380: Our New Autonomous Research Vehicle

Obstacle Avoidance using Learning Model Predictive Control

Obstacle Avoidance using Learning Model Predictive Control