Nature Inspired Design of Autonomous Driving Agent – Realtime Localization, Mapping and Avoidance of Obstacle Based on Motion Parallax

Abstract

We present an approach for nature-inspired design of the driving style of an agent, remotely operating a scale model of a car with obstacle avoidance capabilities. The agent perceives the position of the car from an overhead video camera and conveys its actions to the car via standard radio control transmitter. In order to cope with the video feed latency we propose an anticipatory modeling in which the agent considers its current actions based on the anticipated intrinsic (rather than currently available, outdated) state of the car and its surrounding. Moreover, in a real-time the agent is able (i) to detect a static obstacle with a priori unknown coordinates using onboard video camera, (ii) to map the global position of the obstacle in a nature-inspired way by observing the dynamics of the change of visual angle (i.e., the motion parallax) of the obstacle in several consecutive video frames, and, (iii) in the vicinity of the latter, to employ a potential field-based obstacle avoidance maneuver. Presented work could be seen as a step towards the automated design of the control software of remotely operated vehicles capable to find a safe solution in changeable and uncertain environments.