Radar Based Longitudinal Virtual Bumper Collision Avoidance System Implemented on a Truck

Abstract

In this report, we describe the implementation of the virtual bumper collision avoidance algorithm for highway vehicles.We describe the results from a series of experiments using the virtual bumper collision avoidance algorithm implemented on a Navistar tractor cab. The virtual bumper combines longitudinal and lateral collision avoidance capabilities to control a vehicle in normal and emergency situations. A programmable boundary, the virtual bumper, defines a personal space around the host vehicle. A radar and a laser range sensor were used to sense the location of vehicles in the region in front of the truck. Incursions into the personal space by target vehicles impose a virtual ‘force’ on the host, which in turn modifies the vehicle’s trajectory in order to avoid collisions with objects in the field of view. The virtual bumper longitudinal controller was tested under several driving situations and at several speeds. The experiments included several scenarios: Adaptive Cruise Control, the truck performing a critical stop when the target vehicle ahead is stationary, and situations in which the target suddenly slows down and speeds up and others which simulate stop and go traffic. Results from the virtual bumper longitudinal experiments were favorable. The algorithm demonstrated robustness to sensor noise and the ability to maintain a safe headway for both normal and emergency driving scenarios. We are presently improving the sensing technology and incorporating a road database which contains roadside features in order to greatly reduce, if not eliminate, false target detection.