- Home
- Robot Manipulation
- Videos / Demos
Videos / Demos
Please browse around to see some of our recent work.
Videos and robotic demonstrations are organized from the most to the least recent publication. To find demonstrations related to a particular research, please find it on the sidebar menu.
Online Replanning with Human-in-The-Loop for Non-Prehensile Manipulation
Lead author: Rafael Papallas
Published: Robotics and Automation Letters (RA-L) 2020 (see paper below)
We propose an online-replanning framework based on trajectory-optimization to solve the problems of reaching through clutter and planning uncertainty. The robot fully autonomously tackle these problems until it is expected to fail in which case it falls back to a human operator for help.
Learning Image-Based Receding Horizon Planning for Manipulation in Clutter
Lead author: Wissam Bejjani
Learning discrete and continuous action policies to be used as heuristics for guiding an image-based look-ahead planner in clutter. This system generalizes over different real-world setting and offers transferable manipulation skills to different target objects.
Occlusion-Aware Manipulation: Learning to Act under uncerftainty
Lead author: Wissam Bejjani
When looking at shelf from the side occlusion becomes a major source of uncertainty. Can a recurrent neural model reason over a search and retrieve task under occlusion and physics-based
uncertainties?
Learning Coarse Physics Models
Lead author: Wisdom C. Agboh
Published: CVS 2020 (see paper below)
We build deep neural network coarse physics models for robotic manipulation.
Non-Prehensile Manipulation in Clutter with Human-In-The-Loop
Lead author: Rafael Papallas
Published: ICRA 2020 (see paper below)
We accelerate kinodynamic motion planning in reaching through clutter problems using a human-in-the-loop and sampling-based kinodynamic planners. You can find our ICRA presentation video here.
Human-like Planning for Reaching in Cluttered Environments
Lead author: Mohamed Hasan
Published: ICRA 2020 (see paper below)
We learn high-level planning skills from VR human demonstrations and transfer it to robot planners. Our planner produces scalable high-level plans that can be transferred to any arbitrary robot model.
Manipulation Planning under Changing External Forces
Lead author: Lipeng Chen
Published: Autonomous Robots 2020 (see paper below)
Imagine grasping a wooden board while your friend drills holes into it and cuts pieces off of it. You would predict the forces your friend will apply on the board and choose your grasps accordingly; for example you would rest your palm firmly against the board to hold it stable against the large drilling forces. We developed a manipulation planner to enable a robot grasp objects similarly.
Learning Physics-Based Manipulation in Clutter
Lead author: Wissam Bejjani
Published: IROS 2019 (see paper below)
We project manipulation in clutter task to an abstract representation. The new representation allows for Image-Based Generalization and physics-based look-ahead planning
Combining Coarse and Fine Physics for Manipulation
Lead author: Wisdom C. Agboh
Published: ISRR 2019 (see paper below)
We accelerate physics predictions through parallelization across time, combining coarse and fine physics models.
Task-Adaptive Manipulation
Lead author: Wisdom C. Agboh
Published: WAFR 2018 (see paper below)
Humans adapt their actions, moving fast or slow depending on the task. How can robots exhibit such adaptive manipulation skills?
Planning with a Receding Horizon for Manipulation in Clutter
Lead author: Wissam Bejjani
Published: IEEE Humanoids 2018 (see paper below)
We propose interleaving planning and execution in real-time, in a closed-loop setting, using a Receding Horizon Planner (RHP) for pushing manipulation in clutter. We address the problem of learning a suitable value function based heuristic for RHP.
Online Re-Planning for Real-Time Manipulation
Lead author: Wisdom C. Agboh
Published: IEEE Humanoids 2018 (see paper below)
Real-time manipulation through a novel online stochastic trajectory optimization algorithm. The key is easy parallelization.