A Passive Approach to Autonomous Collision Detection and Avoidance in Uninhabited Aerial Systems.

A new approach to passive detection and avoidance of collision and near-collision with moving obstacles by Uninhabited Aerial Systems (UAS) is proposed in this paper. It takes as inputs the bearings between the ownship and the moving obstacles (intruders) only. Bearings can be measured by passive sensors such as millimetre-wave (mmW) or infra-red (IR) imaging thus adding an important aspect of covertness. The first stage of the approach deals with the detection of the moving obstacle (intruder) and is called collision risk estimator (CRE). The proposed novel approach estimates the risk of collision based on the density of consecutive bearings (constant bearings correspond to high risk and variation in bearings corresponds to low risk). The approach is recursive and thus makes possible to take as many past values of the bearings as practically needed without memorising them. Once a collision is detected, in the second stage of the approach, a decision is taken whether an emergency collision avoidance manoeuvre is necessary or a smoother and optimal (in terms of minimum time) rerouting can take place (de-confliction). The decision which of the two possible alternative avoidance actions is going to be taken can be based on the estimated ‘time-to-collision’ or on the degree of risk and is outside of the scope of the current paper. Emergency collision avoidance manoeuvres are taken in accordance to the Rules of the Air (RoA) and are limited by the ownship kinematic and aerodynamic characteristics only. Optimal re-routing provides a trajectory and the sequence of control actions that are necessary to be taken to ensure this trajectory of the ownship. As a whole, the above stages provide a new approach to passive detection and avoidance (in an optimal fashion) of moving obstacles (intruders) based on bearings only provided by passive sensors.