Mobile Manipulation
Mobile manipulation requires integrating perception, control, motion planning, grasping, navigation and learning in a cohesive manner to act effectively in large unstructured environments. Our approach is to construct a consistent environment representation using visual sensing. Fast motion planners allow the generation of smooth trajectories for the robot. The environment representation is dynamically updated and complemented with information from other types of sensors (tactile, force, etc.). This allows the robot to react quickly to changes in its surroundings. Stiff trajectory controllers allow for tracking the paths effectively while compliant controllers are useful for tasks requiring contact with the environment. Coupled with grasping techniques for different types of objects, this allows the robot to effectively manipulate and move around in different types of environments.
Mobile manipulation requires integrating perception, control, motion planning, grasping, navigation and learning in a cohesive manner to act effectively in large unstructured environments. Our approach is to construct a consistent environment representation using visual sensing. Fast motion planners allow the generation of smooth trajectories for the robot. The environment representation is dynamically updated and complemented with information from other types of sensors (tactile, force, etc.). This allows the robot to react quickly to changes in its surroundings. Stiff trajectory controllers allow for tracking the paths effectively while compliant controllers are useful for tasks requiring contact with the environment. Coupled with grasping techniques for different types of objects, this allows the robot to effectively manipulate and move around in different types of environments.
Integrated Mobile Manipulation
I was part of the team that developed the most advanced mobile manipulation software for unstructured environments. The primary application of this system was in mobile and fixed pick and place scenarios. This system was initially implemented as part of the Arm Navigation and Grasping Pipeline components in ROS for the PR2 robot. It was picked up and implemented on several other robots by other teams, including on industrial robots and the Robonaut. It was further demonstrated in the ICRA 2012 Mobile Manipulation Challenge by a team of Willow Garage researchers, interns and collaborators that I led. This set of capabilities is now implemented in the MoveIt! software.
Related Publications
Demonstrations
I was part of the team that developed the most advanced mobile manipulation software for unstructured environments. The primary application of this system was in mobile and fixed pick and place scenarios. This system was initially implemented as part of the Arm Navigation and Grasping Pipeline components in ROS for the PR2 robot. It was picked up and implemented on several other robots by other teams, including on industrial robots and the Robonaut. It was further demonstrated in the ICRA 2012 Mobile Manipulation Challenge by a team of Willow Garage researchers, interns and collaborators that I led. This set of capabilities is now implemented in the MoveIt! software.
Related Publications
- "Perception, Planning, and Execution for Mobile Manipulation in Unstructured Environments", Chitta, Sachin., Jones, Edward Gil., Ciocarlie, Matei., and Hsiao, Kaijen, IEEE Robotics and Automation Magazine, Special Issue on Mobile Manipulation, Volume 19, Issue 2, 2012 [PDF]
- "Towards Reliable Grasping and Manipulation in Household Environments", Ciocarlie, Matei., Hsiao, Kaijen., Jones, Gil E.., Chitta, Sachin., Rusu, Radu Bogdan., and Sucan, Ioan Alexandru, International Symposium on Experimental Robotics (ISER), New Delhi, India, 2010 [PDF]
- "Contact-Reactive Grasping of Objects with Partial Shape Information", Hsiao, Kaijen., Chitta, Sachin., Ciocarlie, Matei., and Jones, Gil E., IEEE International Conference on Intelligent Robots and Systems (IROS), 2010 [PDF]
Demonstrations
|
"Perception, Planning, and Execution for Mobile Manipulation in Unstructured Environments", Chitta, Sachin., et al, IEEE RAM, Special Issue on Mobile Manipulation.
|
"Pick and Place with an Industrial Robot", ROS-Industrial Consortium, Southwest Research Institute, San Antonio, Texas
|
Tactile and Proprioceptive Sensing
Tactile and/or proprioceptive sensing provides a set of inputs that complement visual sensing very well. These are incredibly powerful tools that we have used for detecting missed or incorrect grasps, slipping of objects in the hand, contact with support surfaces for placing objects and even recognition of the state of liquid containers.
Related Publications
Tactile and/or proprioceptive sensing provides a set of inputs that complement visual sensing very well. These are incredibly powerful tools that we have used for detecting missed or incorrect grasps, slipping of objects in the hand, contact with support surfaces for placing objects and even recognition of the state of liquid containers.
Related Publications
- "Human-Inspired Robotic Grasp Control with Tactile Sensing", Romano, Joseph., Hsiao, Kaijen., Niemeyer, Günter., Chitta, Sachin., and Kuchenbecker, Katharine J., IEEE Transactions on Robotics, Volume 27, Issue 6, p.1067-1079, 2011 [PDF]
- "Tactile Sensing for Mobile Manipulation", Chitta, Sachin., Sturm, Jürgen., Piccoli, Matthew., and Burgard, Wolfram, IEEE Transactions on Robotics, 2011 [PDF]
- "Tactile Object Class and Internal State Recognition for Mobile Manipulation", Chitta, Sachin., Piccoli, Matthew., and Sturm, Jürgen, IEEE International Conference on Robotics and Automation, Anchorage, Alaska, 2010 [PDF]
|
Joe Romano's Willow Garage intern Video
|
Navigation
2D mobile base navigation in unstructured environments is fairly well-understood and has been addressed using a 2D projection of the world (and the robot). However, navigation gets harder when the robot is carrying or pushing an object since it needs to now account for the size and position of the object. In some cases, this requires a full 3D model of the robot and the world. We extended autonomous mobile base navigation to 3D, accounting for the full 3D model of the world and the 3D representation of the robot. We demonstrated our approach in two situations:
2D mobile base navigation in unstructured environments is fairly well-understood and has been addressed using a 2D projection of the world (and the robot). However, navigation gets harder when the robot is carrying or pushing an object since it needs to now account for the size and position of the object. In some cases, this requires a full 3D model of the robot and the world. We extended autonomous mobile base navigation to 3D, accounting for the full 3D model of the world and the 3D representation of the robot. We demonstrated our approach in two situations:
- mobile robot navigation while pushing a cart
- mobile robot navigation while carrying a large object - In this case we used a clever optimization for collision checking to minimize the number of 3D collision checks. This allowed the PR2 robot to carry large objects like a basket in unstructured environments, using the 2D collision representation of the environment as much as possible while doing 3D collision checks only when absolutely necessary.
- "Navigation in Three-Dimensional Cluttered Environments for Mobile Manipulation", Hornung, Armin., Phillips, Mike., Jones, Edward Gil., Bennewitz, Maren., Likhachev, Maxim., and Chitta, Sachin, IEEE International Conference on Robotics and Automation, Minneapolis, Minnesota, 2012 [PDF]
- "Cart Pushing with a Mobile Manipulation System: Towards Navigation with Moveable Objects", Scholz, Jonathan., Chitta, Sachin., Marthi, Bhaskara., and Likhachev, Maxim International Conference on Robotics and Automation, Shanghai, China, 2011 [PDF]
|
|
|
Applications: Door Opening
I have worked on several door opening systems with the PR2 robot using a combination of compliant control and motion planning techniques. This includes opening spring loaded doors. The most comprehensive exploration was for the "Milestone 2" set of tasks for opening 10 doors consecutively with the PR2 robot.
Related Publications
I have worked on several door opening systems with the PR2 robot using a combination of compliant control and motion planning techniques. This includes opening spring loaded doors. The most comprehensive exploration was for the "Milestone 2" set of tasks for opening 10 doors consecutively with the PR2 robot.
Related Publications
- "A Single Planner for a Composite Task of Approaching, Opening and Navigating through Non-spring and Spring-loaded Door", Gray, Steven., Chitta, Sachin., Kumar, Vijay., and Likhachev, Maxim, IEEE International Conference on Robotics and Automation, Karlsruhe, Germany, 2013 [PDF]
- "Autonomous Door Opening and Plugging In with a Personal Robot", Meeussen, Wim., Wise, Melonee., Glaser, Stuart., Chitta, Sachin., McGann, Conor., Mihelich, Patrick., Marder-Eppstein, Eitan., Muja, Marius., Eruhimov, Victor., Foote, Tully, et al., IEEE International Conference on Robotics and Automation, Anchorage, Alaska, 2010 [PDF]
- "Planning for Autonomous Door Opening with a Mobile Manipulator", Chitta, Sachin., Cohen, Benjamin., and Likhachev, Maxim, IEEE International Conference on Robotics and Automation, Anchorage, Alaska, 2010 [PDF]
- "Milestone 2: Autonomous Door Opening and Plugging In with a Personal Robot"
|
|
