My research focuses on robotics manipulation. In particular, I am interested in developing algorithms for non-prehensile, physics-based manipulation in clutter in conjunction with some human guidance. Imagine a robot trying to reach for an object from a warehouse shelf or from the back of a cluttered fridge. My research deals with this kind of challenging problems where the robot needs to reason about how and which objects to push out of the way to make space to reach for a goal object.
- UR5 Controller, an open-source OpenRAVE controller for UR5 robot and a singularity container for it.
- A UR5 model for MuJoCo physics-simulator.
- An OpenRAVE environment for the Amazon Picking Challenge competition.
- Maintaining or_tf an OpenRAVE plugin for real-time object tracking from a Motion-Capture system.
- AFHEA (HEA Associate Fellow)
At the very bottom, there is a list of my publications. They are also available on my Google Scholar website here.
Imagine a robot at home trying to reach for a bottle of water from the back of a cluttered fridge for a disabled person. The research community has mainly focused on prehensile manipulation and on approaches that treat the environment as a static one that prevents the robot from interacting with other objects. The real-world, however, is governed by physics and if we want robots to get into our environments and become useful we need to make them physics aware and allow them to interact with other objects.
My research focuses on non-prehensile physics-based manipulation in clutter and deals with this kind of challenging problems where the robot needs to reason how and which objects to push out of the way to make space to reach for a goal object in a cluttered environment like a fridge or a shelf. This problem is extremely hard for today's robots and autonomous solutions are suffering from long planning times and low success rates for hard instances of the problem. We recognise that humans are experts in this context and we have a great intuition of how to perform these actions in a split of a second, yet we cannot explain how we can make these decisions. My research focuses on Human-In-The-Loop systems and leverages the human intuition to accelerate the performance of robots in manipulating cluttered environments with minimal human effort.
We propose a human-operator guided planning approach to pushing-based manipulation in clutter. Most recent approaches to manipulation in clutter employs randomized planning. The problem, however, remains a challenging one where the planning times are still in the order of tens of seconds or minutes, and the success rates are low for difficult instances of the problem. We build on these control-based randomized planning approaches, but we investigate using them in conjunction with human-operator input. In our framework, the human operator supplies a high-level plan, in the form of an ordered sequence of objects and their approximate goal positions. We present experiments in simulation and on a real robotic setup, where we compare the success rate and planning times of our human-in-the-loop approach with fully autonomous sampling-based planners. We show that with a minimal amount of human input, the low-level planner can solve the problem faster and with higher success rates.
- BSc (Honours) Computer Science (Industrial), The University of Leeds