Challenges for AI: Mobile Robots on Construction Sites. Tim Detert

Transcription

1 Challenges for AI: Mobile Robots on Construction Sites Tim Detert

2 Challenges for AI: Mobile Robots on Construction Sites What will Mobile Robots on Construction Sites be Like? In the future it eventually may take only 2 percent of the population, coupled with some pretty remarkable automation, to accomplish what it currently takes 90 percent of our population to accomplish. Troy K. Schneider 01/2016 [ Are we thinking about artificial intelligence all wrong?] [Disney UK Folie 2 / 14

3 Mobile Robots on Construction Sites The Endless Wall introduced by Gramazio & Kohler (ETH Zürich) in 2011 building an architectural complex endless brick wall direct employment of industrial robots on construction sites innovative scanning system for self-orientation [ [ responding to construction tolerances and adapting to changing conditions Folie 3 / 14

4 Mobile Robots on Construction Sites The Worlds First Mobile Drilling Robot is LIVE released by nlink in 02/2015 mounting of equipment to the ceiling fire sprinkler systems lighting fixtures reducing strain on workers based on building information files combining cameras and lasers for positioning [ Folie 5 / 14

5 Challenge: Mobile Robots on Construction Sites mobility [ [ [ collaboration [ [ [ perception [ [ [ Folie 6 / 14

6 Developing a Robotic System for Asbestos Removal launched in 02/2016 as a H2020 Innovation Action clearance and refurbishment of rehabilitation sites asbestos removal in hazardous environments tremendous market potential simplified environment allows semi-autonomous operation removal tasks with high repetition and low complexity adapted design and technology [ Folie 7 / 14

7 International Consortium of seven Partners launched in 02/2016 as a H2020 Innovation Action clearance and refurbishment of rehabilitation sites asbestos removal in hazardous environments tremendous market potential simplified environment allows semi-autonomous operation removal tasks with high repetition and low complexity adapted design and technology [ Folie 8 / 14

8 Automation of Asbestos Removal Tasks Ground Walls Ceilings Tools Asbestos in flats Plastic tiles and glue Ceramic tiles Plastering Paint Folie 9 / 14

9 Geometric and Task-Specific Boundary Conditions mobility collaboration perception limited mass of 300kg due to allowed load on the floor small 0.6x0.8m footprint and compact folding for narrow corridors differential drive and OmniWheels for fragile floor seclusion specific kinematics for reachability of floor/ceiling situation prediction to adapt footprint adaptive stabilization to account for processing forces and the small support polygon Folie 10 / 14

10 Collaboration of Operator and Mobile Robotic Units mobility perception collaboration user input for task allocation via semantic mapping asbestos identification task definition and parameter specification process supervision collaboration of robotic units in a shared workspace autonomous task planning and sharing collision avoidance and trajectory generation Folie 11 / 14

11 Navigation and Process Surveilance mobility perception collaboration low visibility and dust in changing environments local process monitoring using 1-D radar technology Self-Localization and Mapping using sensor data fusion for laser, radar and camera data context dependency of information significance and confidence of data weighting factors and decision logics Folie 12 / 14

12 Artificial Intelligence might help on our way mobility [ collaboration perception 2016 tele-operation 2017 partial autonomy 2019 task-autonomy and collaboration Folie 13 / 14

13 Thank you for your kind attention Contact: Tim Detert Department of Mechanism Theory and Dynamics of Machines Kackertstraße Aachen (+49)