Collision Avoidance Systems for Manual-assisted Driving

Abstract:
The paper describes the design and implementation of safe driving-modes for industrial vehicles, by means of manual-assisted driving, in which both the driver and an "intelligent" robot co-exist. In this context, manual-assisted driving can be defined as a way of driving where drivers play their usual role and the control system just prevents eventual collisions (that is a semi-autonomous mobile robot system). In this sense, the system reacts by altering driver commands in such a way that the collision is avoided. The paper also addresses the problem of evaluating this kind of driving mode, comparing it with others approaches, such as AGVs with different degrees of autonomy. To achieve this goal a methodology and benchmark tools are defined to grade the performance of each solution. The proposed benchmark requires to run different experiments at different velocities and different scenarios covering from manual driving to autonomous navigation, where the hybrid solution proposed here is of course also considered.