Soft/Tactile perception and control for robust manipulation in homes
Manipulation in domestic environments is a hard problem due to multiple challenges such as the need to cope with clutter, lighting variations, safety requirements etc. From the task standpoint, manipulators must be capable of stable grasps, precise placement and safe interactions during inadvertent contact.
While recent years have seen tremendous progress in camera perception powered by advanced in machine learning, reliance on vision alone as the primary sensing modality may be insufficient. Furthermore, given the inherent mechanics of contact, it is important to understand and exploit the mechanical properties of the robot manipulator hardware itself.
Thus, towards achieving truly robust manipulation capability in homes, two research themes that hold a lot of potential are (i) tactile manipulation and (ii) soft robotics. In this project, we jointly explore these themes through our Soft-bubble visuotactile (camera-based tactile) sensors and end-effectors. There are broadly 2 research thrusts exploring these themes: (i) development of newer and more effective soft and perceptive tactile sensors, and (ii) developing algorithms for the perception and control using tactile feedback and exploiting softness.
This project has resulted so far in several publications, patent submissions and 1 best paper award.
Sample Publications
- Kuppuswamy, N., Alspach, A., Uttamchandani, A., Creasey, S., Ikeda, T., & Tedrake, R. (2020). Soft-Bubble grippers for robust and perceptive manipulation. arXiv preprint arXiv:2004.03691. (Arxiv link)
- Kuppuswamy, N., Castro, A., Phillips-Grafflin, C., Alspach, A., & Tedrake, R. (2019). Fast model-based contact patch and pose estimation for highly deformable dense-geometry tactile sensors. IEEE Robotics and Automation Letters, 5(2), 1811-1818. [IEEE Robotics and Automation Letters Best paper award (2019)] (IEEE link / open access)
- Ambrus, R., Guizilini, V., Kuppuswamy, N., Beaulieu, A., Gaidon, A., & Alspach, A. Monocular Depth Estimation for Soft Visuotactile Sensors. arXiv preprint arXiv:2101.01677. (Arxiv link)
- Alspach, A., Hashimoto, K., Kuppuswamy, N., & Tedrake, R. (2019, April). Soft-bubble: A highly compliant dense geometry tactile sensor for robot manipulation. In 2019 2nd IEEE International Conference on Soft Robotics (RoboSoft) (pp. 597-604). IEEE. (IEEE link) (Download link)
Naveen Kuppuswamy 