Motion Strategies for Visibility based Target Tracking in Unknown Environments

Abstract

Target tracking is an interesting problem and has important applications in security and surveillance systems, personal robotics, computer graphics, and many other domains. The focus of this thesis is on computing motion strategies to keep a moving target in view in a dynamic and unknown environment using visual sensors. The problem of motion planning is complicated by the mobility and visual obstructions from the obstacles in the environment. Without using a-priori information about the target and the environment, this thesis proposes an online tracking algorithm which plans its motion strategy using local information from on-board sensors. In order to track intelligently, the tracker has to choose an action which lowers the danger of losing the target in the future while maintaining it under view in the current step. This thesis proposes a measure called relative vantage which combines the risk of losing the target in the current time to the risk of losing the target in the future. A local greedy tracking algorithm called vantage tracker is proposed which chooses actions to minimize this risk measure.

Implementing a robust robotic tracker requires dealing with sensing limitations such as maximum range, field-of-view limits, motion limitations such as maximum speed bound, non-holonomic constraints and operational limitations such as obstacle avoidance, stealth, e.t.c. This thesis proposes a general tracking framework that incorporates these limitations into the problem of online target tracking. A real robotic tracker was setup using a simple laser range finder and a differential drive robot base and the hardware limitations were addressed in the tracking framework as planning constraints.

Such a tracker was able to successfully follow a person in a crowded environment. A stealth constraint was formulated where the tracker has to maintain sight of the target while trying to avoid being detected. Incorporating this stealth constraint into the tracking problem, a stealth tracking algorithm was developed and analyzed for various environments in simulation.

In a 3-D environment, the visibility relationships become complex easily. Moreover, the additional dimension available to the target makes the tracking problem more difficult. A 3-D vantage tracker was developed by generalizing the approach pertaining to the 2-D tracker. Such a tracker generates intelligent tracking actions by exploiting the additional dimension. As an example a robotic helicopter generates a vertical motion to avoid occlusion of the target due to the buildings in an urban scenario when it can improve its visibility by doing so. Such a behavior was generated based only on the locally sensed geometric parameters and no a priori knowledge of the layout or the model of the obstacles in the environment was used.

Extensive simulation and hardware results show consistently the improvement in tracking performance of the vantage tracker based tracking framework both in 2-D and in 3-D as compared to previous approaches such as visual servo and those based on increasing the shortest distance to escape for the target.