2025 Theses Doctoral

Robot Learning with Sparsity and Scarcity

Xu, Jingxi

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 challenge into (1) data sparsity from the angle of data representation and (2) data scarcity from the angle of data quantity. The data sparsity problem means that there is a large proportion of empty space or non-relevant information in the data we collected. Robotics is the science of interaction. We have a piece of software or an algorithm embodied inside a piece of hardware, and then the robot needs to interact actively with the environment to collect useful information. The sequential manner of such interaction and the lapse between two consecutive actions make robotic data inherently very sparse. On the other hand, the data scarcity issue is that the sheer amount of data we can collect in the domain of interest is very limited. In contrast to the richness and accessibility of text, image, and video data available on the Internet, it is extremely difficult to collect data from physical hardware or humans on a large scale.

In this thesis, I will discuss my PhD work on two selected domains: (1) tactile manipulation 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. I will discuss my 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. I will discuss my 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. My work develops machine learning algorithms that enable intent inferral with minimal data, including semi-supervised, meta-learning, reciprocal learning, and generative AI methods.