Learning from Experience for Rapid Generation of Local Car Maneuvers

TL;DR

This paper introduces a deep neural network trained with a novel weakly supervised method to rapidly generate feasible, smooth, and nearly optimal local paths for autonomous vehicles, significantly outperforming existing planners in success rate.

Contribution

The paper presents a new weakly supervised training approach for a DNN that enables fast, feasible local path planning for vehicles, improving success rates and maintaining path quality.

Findings

Path generation time ~40 ms

Outperforms existing planners in success rate

Generated paths are smooth and comparable to conventional methods

Abstract

Being able to rapidly respond to the changing scenes and traffic situations by generating feasible local paths is of pivotal importance for car autonomy. We propose to train a deep neural network (DNN) to plan feasible and nearly-optimal paths for kinematically constrained vehicles in small constant time. Our DNN model is trained using a novel weakly supervised approach and a gradient-based policy search. On real and simulated scenes and a large set of local planning problems, we demonstrate that our approach outperforms the existing planners with respect to the number of successfully completed tasks. While the path generation time is about 40 ms, the generated paths are smooth and comparable to those obtained from conventional path planners.