ABSTRACT
With recent advances in hardware, sensing, and algorithms, we are witnessing the emergence of a new robotics industry. I will present a few examples of new services provided by upcoming service robots. With the introduction of new service robots in diverse domains, we can expect that more service robots will be assisting us in the near future in places, such as offices, malls, and homes. But, for a robot to coexist with humans and operate successfully in crowded and dynamic environments, a robot must be able to learn from experiences to act safely and harmoniously with human participants in the environment. I will discuss research challenges for service robots and our attempts to address those challenges.
In particular, I will present our recent work on leveraged Gaussian process regression for learning with counterexamples to enhance safety and social acceptability of service robots. While existing learning from demonstration algorithms assume that demonstrations are collected from experts, the proposed method alleviates such assumption by allowing demonstrations from casual or novice users. To learn from demonstrations of mixed quality, we present a leverage optimization method using the correlation structure in leveraged Gaussian processes. I will also present how the same concept can be applied to inverse reinforcement learning and scaled up using deep learning.