Blog

What if robots could pick up any object without needing detailed information about its shape or size? Traditionally, robotic grasping has relied on precise object models, but newer methods are shifting towards flexibility and adaptability. In our work, we applied Bayesian Optimization (BO) to a robotic hand equipped with tactile sensors, to learn from previous trials using tactile feedback and grasp diverse objects safely and robustly. We have also developed new heuristics that improve the system’s efficiency by reducing the number of trials needed for successful grasps and make it more effective in real-world applications.

By combining tactile feedback and BO, we’re creating robotic systems that can adapt quickly and handle a wide range of tasks, all while minimizing the need for extensive data. To know more in detail, please follow the link to the article and IntelliMan project.

Join it at the link:
https://ieeexplore.ieee.org/document/10706823