Improving Automated Driving through POMDP Planning with Human Internal States

TL;DR

This paper demonstrates that POMDP planning incorporating human internal states enhances safety and efficiency in autonomous freeway lane changes, outperforming traditional MDP approaches in simulated scenarios.

Contribution

It introduces a POMDP-based approach with human internal states for autonomous driving, showing significant safety improvements over MDP baselines.

Findings

POMCPOW reduces unsafe situations by half.

POMCPOW increases success rate by 50%.

Outperforms MDP and matches or exceeds QMDP performance.

Abstract

This work examines the hypothesis that partially observable Markov decision process (POMDP) planning with human driver internal states can significantly improve both safety and efficiency in autonomous freeway driving. We evaluate this hypothesis in a simulated scenario where an autonomous car must safely perform three lane changes in rapid succession. Approximate POMDP solutions are obtained through the partially observable Monte Carlo planning with observation widening (POMCPOW) algorithm. This approach outperforms over-confident and conservative MDP baselines and matches or outperforms QMDP. Relative to the MDP baselines, POMCPOW typically cuts the rate of unsafe situations in half or increases the success rate by 50%.