Map Optimization with Distance-Based Covariance in Industrial Field
2018 IEEE 8TH ANNUAL INTERNATIONAL CONFERENCE ON CYBER TECHNOLOGY IN AUTOMATION, CONTROL, AND INTELLIGENT SYSTEMS (IEEE-CYBER)(2018)
摘要
Precise and reliable LIDAR based mapping and localization are critical technologies to automatic guided vehicles (AGV) in flexible industrial applications. In the complex dynamic industrial environment, highly reflective and stationary landmarks provide much reliability and less computational cost than using point cloud. However, the sparse map only based on landmark contains fewer constraints, and the accuracy of each measurement reduces with increasing range. Hence, scan matching accumulates much error over distance. To improve map precision, we present the approach using graph optimization with distance-based covariance for LIDAR mapping. The error model of LIDAR detection over distance is built and used for scaling the covariance of sensor measurements, and then graph-based optimization of constraint networks is applied with the covariance. We provide experiment results which illustrate the effectiveness of the proposed approach.
更多查看译文
关键词
LIDAR detection,map optimization,automatic guided vehicles,flexible industrial applications,complex dynamic industrial environment,reflective landmarks,stationary landmarks,LIDAR mapping,graph-based optimization,distance-based covariance analysis,AGV,reliability,sensor measurements
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络