Uploaded October 2025 | Updated September 2026, 9 hours ago
Unlike in language or vision, one of the fundamental challenges in robot learning is the lack of access to vast data resources. We can further break down the problem into (1) data sparsity from the angle of data representation and (2) data scarcity from the angle of data quantity. In this talk, Jingxi discusses selected works on two domains: (1) tactile sensing and (2) rehabilitation robots, which are exemplars of data sparsity and scarcity, respectively. Tactile sensing is an essential modality for robotics, but tactile data are often sparse, and for each interaction with the physical world, tactile sensors can only obtain information about the local area of contact. He discussed his work on learning vision-free tactile-only exploration and manipulation policies through model-free reinforcement learning to make efficient use of sparse tactile information. On the other hand, rehabilitation robots are an example of data scarcity to the extreme due to the significant
challenge of collecting biosignals from disabled-bodied subjects at scale for training. He discusses his work in collaboration with the medical school and clinicians on intent inferral for stroke survivors, where a hand orthosis developed in our lab collects a set of biosignals from the patient and uses them to infer the activity that the patient intends to perform, so the orthosis can provide the right type of physical assistance at the right moment. His work develops machine learning algorithms that enable intent inferral with minimal data, including semi-supervised, meta-learning, and generative AI methods.
Jingxi Xu is a research scientist at the RAI Institute (formerly known as the Boston Dynamics AI Institute). He recently graduated with a Ph.D. in Computer Science at Columbia University, co-advised by Professor Matei Ciocarlie and Shuran Song. He received his bachelor’s degree from The University of Edinburgh, with First Class Honours and Class Medal (overall grade ranks first in class), and he received an MS from Columbia, where he worked closely with Professor Peter Allen. At Columbia, he is a recipient of the CA fellowship and the Andrew P. Kosoresow Memorial Award for Excellence in Teaching and Service. He also spent some time as a research intern at the RAI Institute and as a visiting researcher at MIT and the University of Pennsylvania (GRASP Lab). His research focuses on robot learning, and he has published at many robotics conferences/journals, including RA-L, ICRA, CoRL, IROS, AuRo, CASE, L4DC, VR, and BioRob, including a best paper finalist award at BioRob 2022.
Unlike in language or vision, one of the fundamental challenges in robot learning is the lack of access to vast data resources. We can further break down the problem into (1) data sparsity from the angle of data representation and (2) data scarcity from the angle of data quantity. In this talk, Jingxi discusses selected works on two domains: (1) tactile sensing and (2) rehabilitation robots, which are exemplars of data sparsity and scarcity, respectively. Tactile sensing is an essential modality for robotics, but tactile data are often sparse, and for each interaction with the physical world, tactile sensors can only obtain information about the local area of contact. He discussed his work on learning vision-free tactile-only exploration and manipulation policies through model-free reinforcement learning to make efficient use of sparse tactile information. On the other hand, rehabilitation robots are an example of data scarcity to the extreme due to the significant
challenge of collecting biosignals from disabled-bodied subjects at scale for training. He discusses his work in collaboration with the medical school and clinicians on intent inferral for stroke survivors, where a hand orthosis developed in our lab collects a set of biosignals from the patient and uses them to infer the activity that the patient intends to perform, so the orthosis can provide the right type of physical assistance at the right moment. His work develops machine learning algorithms that enable intent inferral with minimal data, including semi-supervised, meta-learning, and generative AI methods.
Jingxi Xu is a research scientist at the RAI Institute (formerly known as the Boston Dynamics AI Institute). He recently graduated with a Ph.D. in Computer Science at Columbia University, co-advised by Professor Matei Ciocarlie and Shuran Song. He received his bachelor’s degree from The University of Edinburgh, with First Class Honours and Class Medal (overall grade ranks first in class), and he received an MS from Columbia, where he worked closely with Professor Peter Allen. At Columbia, he is a recipient of the CA fellowship and the Andrew P. Kosoresow Memorial Award for Excellence in Teaching and Service. He also spent some time as a research intern at the RAI Institute and as a visiting researcher at MIT and the University of Pennsylvania (GRASP Lab). His research focuses on robot learning, and he has published at many robotics conferences/journals, including RA-L, ICRA, CoRL, IROS, AuRo, CASE, L4DC, VR, and BioRob, including a best paper finalist award at BioRob 2022.
