simultaneous localization and mapping VII - seamless loop closing

This video shows laser scanner based Simultaneous Localization And Mapping (SLAM). What you see is a slow moving electric vehicle in an outdoor environment. The vehicle is equipped only with two ibeo laser scanners and nothing else since this particular SLAM technique requires no other sensors nor requires any mathematical model of the vehicle. The algorithm works in real time and on any type of platform, including humans or humanoid robots. This video shows version 4.0 of the algorithm which is accurate enough to seamlessly close loops of hundreds of meters, something less accurate versions shown in previous videos would not do so well. In this video the algorithm keeps in memory all the laser points from the start, other videos would use a time window which makes more sense for long drive applications but what I want to show here is loop closing which implies to keep all the points in order to show that no position error is apparent when passing a second time at the same location. In this video we can see too the instant laser points in order to show how noisy is the data we are working with.

Category:

Science & Technology

Tags:

• robot
• outdoor
• simultaneous
• localization
• mapping
• seamless
• loop
• closing
• exploration
• laser
• scanner
• lidar
• ibeo
• hokuyo
• autonomous
• navigation
• vehicles
• mobile
• robotics
• slam
• survey
• SRI
• karto
• SDK