基于深度优先搜索分支定界法的Graph-SLAM后端优化算法改进

Improvement of Graph-SLAM Back-End Optimization Algorithm Based on Depth-Priority Search Branch and Bound

针对未知环境中机器人定位的问题,提出了一种深度优先搜索分支定界法的优化改进算法。graph-slam是一种离线slam方法 ,通过采用该优化改进算法可以使graph-slam的后端优化所需耗时减少,使整个系统的效率提高,使其能基本达到一个实时的效果。实验结果表明,该优化改进算法能够使系统运行效率提高近50%,同时能保证系统的稳定性和精确度的要求。

Aiming at the problem of robot localization in unknown environment, an improved algorithm of depth-priority search branch and bound method is proposed. Graph-slam is an off-line slam method. By using this optimization algorithm, graph-slam’s back-end optimization can be reduced by time-consuming, so that the effciency of the whole system can be improved to achieve a real-time effect. The experimental results show that the improved algorithm can improve the running effciency of the system by nearly 50%, and can guarantee the stability and accuracy of the system.