Workshop contents: We will review fundamentals of SLAM, and provide some details on the popular computational methods in the front-end and the back-end, such as feature tracking, scan matching, pose graph optimization, and bundle adjustment. We will then discuss shortcoming of current SLAM methods and trends.

We will then present our outdoor data-set (including buildings, trees, fences, construction material, uneven terrain,…), and discuss in details the results of our experiments with popular Lidar and visual SLAM implementations.

Time: Fri Oct. 8th, 13:00-16:00

Duration: 3x45min

Preliminary plan:

13:00 -- 13:10 Introduction to AI Hub Tampere and Autonomous Mobile Machines research

13:10 -- 13:45 SLAM fundamentals, visual SLAM and Lidar SLAM

15min break

14:00 -- 14:20 SLAM theory continues

14:20 -- 14:45 Results and our experiences with SLAM; Introduction to test environment and data set

15min break (and time to finalize the environment installations)

15:00 -- 15:45 Results discussion and try yourself

Target audience: Mobile machines automation engineers and practitioners

Form: Zoom. The link will be sent to registered participants' emails prior to the event.

Required knowledge and preparation for the workshop: We will provide one data-set and an implementation in ROS environment, so you can try yourself to explore the results. We will provide the code example, instruction about the ROS environment, and data before the workshop.

Required pre-knowledge and/or software: ROS

Speakers: Bharath Garigipati, Reza Ghabcheloo (Associate Professor)