The ability of sensors to self-organize is an important asset in surveillance sensor networks. Self-organize implies self-control at the sensor level and coordination at the network level. Biologically inspired approaches have recently gained significant attention as a tool to address the issue of sensor control and coordination in sensor networks. These approaches are exemplified by the two well-known algorithms, namely, the Flocking algorithm and the Anti-Flocking algorithm. Generally speaking, although these two biologically inspired algorithms have demonstrated promising performance, they expose deficiencies when it comes to their ability to maintain simultaneous robust dynamic area coverage and target coverage. These two coverage performance objectives are inherently conflicting. This paper presents Semi-Flocking, a biologically inspired algorithm that benefits from key characteristics of both the Flocking and Anti-Flocking algorithms. The Semi-Flocking algorithm approaches the problem by assigning a small flock of sensors to each target, while at the same time leaving some sensors free to explore the environment. This allows the algorithm to strike balance between robust area coverage and target coverage. Such balance is facilitated via flock-sensor coordination. The performance of the proposed Semi-Flocking algorithm is examined and compared with other two flocking-based algorithms once using randomly moving targets and once using a standard walking pedestrian dataset. The results of both experiments show that the Semi-Flocking algorithm outperforms both the Flocking algorithm and the Anti-Flocking algorithm with respect to the area of coverage and the target coverage objectives. Furthermore, the results show that the proposed algorithm demonstrates shorter target detection time and fewer undetected targets than the other two flocking-based algorithms.

This paper introduces a hybrid algorithm of deep reinforcement learning (RL) and Force-based motion planning (FMP) to solve distributed motion planning problem in dense and dynamic environments. Individually, RL and FMP algorithms each have their own limitations. FMP is not able to produce time-optimal paths and existing RL solutions are not able to produce collision-free paths in dense environments. Therefore, we first tried improving the performance of recent RL approaches by introducing a new reward function that not only eliminates the requirement of a pre supervised learning (SL) step but also decreases the chance of collision in crowded environments. That improved things, but there were still a lot of failure cases. So, we developed a hybrid approach to leverage the simpler FMP approach in stuck, simple and high-risk cases, and continue using RL for normal cases in which FMP can't produce optimal path. Also, we extend GA3C-CADRL algorithm to 3D environment. Simulation results show that the proposed algorithm outperforms both deep RL and FMP algorithms and produces up to 50% more successful scenarios than deep RL and up to 75% less extra time to reach goal than FMP.

No abstract

No abstract