Three papers from CPSLAB are accepted to IROS 2017

[2017.06.15]

Following three papers are accepted to the IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2017):

  • Online Learning to Approach a Person with No-Regret by Hyemin Ahn, Yoonseon Oh, Sungjoon Choi, Claire J. Tomlin, and Songhwai Oh
    • Abstract: Each person has a different personal space and behaves differently when another person approaches. Based on this observation, we propose a novel method to learn how to approach a person comfortably based on the person’s preference while avoiding uncomfortable encounters. We propose a personal comfort field to learn each person’s preference about an approaching object. A personal comfort field is based on existing theories in anthropology and personalized for each user through repeated encounters. We propose an online method to learn a personal comfort field of a user, i.e., personalized learning, based on the concept from Gaussian process upper confidence bound and show that the proposed method has no-regret asymptotically. The effectiveness of the proposed method has been extensively validated in simulation and real-world experiments. Results show that the proposed method can gradually learn the personalized approaching behavior preferred by a user as the number of encounters increases.
    • Video
  • Cost-Aware Path Planning under Co-Safe Temporal Logic Specifications by Kyunghoon Cho, Junghun Suh, Claire J. Tomlin, and Songhwai Oh
    • Abstract: This paper presents a path planning algorithm for generating a cost-efficient path which satisfies mission requirements specified in linear temporal logic (LTL). We assume that a cost function is defined over the configuration space. Examples of a cost function include hazard levels, wireless connectivity, and energy consumption, to name a few. The proposed method consists of two parts: (1) sampling-based cost-aware path planning considering the vehicle dynamics based on RRT, and (2) a high-level logic which determines how to extend the RRT tree based on the spatio-temporal specifications of an LTL formula. In order to find a low-cost trajectory with computational efficiency, the proposed method expand the RRT tree with long extensions using the cross entropy optimization method, while the rewiring step of RRT is used to preserve the asymptotic optimality. In simulation and experiments, we show that the proposed method performs favorably compared to existing methods.
    • Video
  • Scalable Robust Learning from Demonstration with Leveraged Deep Neural Networks by Sungjoon Choi, Kyungjae Lee, and Songhwai Oh
    • Abstract: In this paper, we propose a novel algorithm for learning from demonstration, which can learn a policy function robustly from a large number of demonstrations with mixed qualities. While most of the existing approaches assume that demonstrations are collected from skillful experts, the proposed method alleviates such restrictions by estimating the proficiency level of each demonstration using leverage optimization. A novel leveraged cost function is proposed to represent a policy function using deep neural networks by reformulating the objective function of leveraged Gaussian process regression using the representer theorem. The proposed method is successfully applied to autonomous track driving tasks, where a large number of demonstrations with mixed qualities are given as training data without labels.
    • Video