Acquiring Accurate Human Responses to Robots’ Questions
In task-oriented robot domains, a human is often designated as a supervisor to monitor the robot and correct its inferences about its state during execution. However, supervision is expensive in terms of human effort. Instead, we are interested in robots asking non-supervisors in the environment for state inference help. The challenge with asking non-supervisors for help is that they may not always understand the robot’s state or question and may respond inaccurately as a result. We identify four different types of state information that a robot can include to ground non-supervisors when it requests help—namely context around the robot, the inferred state prediction, prediction uncertainty, and feedback about the sensors used for the predicting the robot’s state. We contribute two wizard-of-oz’d user studies to test which combination of this state information increases the accuracy of non-supervisors’ responses. In the first study, we consider a block-construction task and use a toy robot to study questions regarding shape recognition. In the second study, we use our real mobile robot to study questions regarding localization. In both studies, we identify the same combination of information that increases the accuracy of responses the most. We validate that our combination results in more accurate responses than a combination that a set of HRI experts predicted would be best. Finally, we discuss the appropriateness of our found best combination of information to other task-driven robots.
KeywordsHuman-robot interaction Asking for help User studies
Unable to display preview. Download preview PDF.
- 4.Asoh H, Hayamizu S, Hara I, Motomura Y, Akaho S, Matsui T (1997) Socially embedded learning of the office-conversant mobile robot jijo-2. In: 15th international joint conference on artificial intelligence, pp 880–885 Google Scholar
- 5.Asoh H, Motomura Y, Hara I, Akaho S, Hayamizu S, Matsui T (1996) Acquiring a probabilistic map with dialogue-based learning. In: Proceedings of ROBOLEARN-96, pp 11–18 Google Scholar
- 6.Banbury S, Seldcon S, Endsley M, Gordon T, Tatlock K (1998) Being certain about uncertainty: How the representation of system reliability affects pilot decision making. In: Human factors and ergonomics society 42nd annual meeting Google Scholar
- 8.Biswas J, Veloso M (2010) Wifi localization and navigation for autonomous indoor mobile robots. In: ICRA 2010, pp 4379–4384 Google Scholar
- 11.Cohn D, Atlas L, Ladner R (1994) Improving generalization with active learning. Mach Learn 15(2):201–221 Google Scholar
- 15.Faulring A, Myers B, Mohnkern K, Schmerl B, Steinfeld A, Zimmerman J, Smailagic A, Hansen J, Siewiorek D (2010) Agent-assisted task management that reduces email overload. In: IUI’10: proceeding of the 14th international conference on intelligent user interfaces, pp 61–70 Google Scholar
- 18.Horvitz E (1999) Principles of mixed-initiative user interfaces. In: CHI’99: proceedings of the SIGCHI conference on Human factors in computing systems, pp 159–166 Google Scholar
- 21.Mcnee S, Lam SK, Guetzlaff C, Konstan JA, Riedl J (2003) Confidence displays and training in recommender systems. In: Proceedings of the 9th IFIP TC13 international conference on humancomputer interaction (INTERACT). IOS Press, Amsterdam, pp 176–183 Google Scholar
- 24.Rosenthal S, Biswas J, Veloso M (2010) An effective personal mobile robot agent through a symbiotic human-robot interaction. In: AAMAS’10: 9th international joint conference on autonomous agents and multiagent systems, pp 915–922 Google Scholar
- 25.Rosenthal S, Dey AK, Veloso M (2009) How robots’ questions affect the accuracy of the human responses. In: The international symposium on robot-human interactive communication, pp 1137–1142 Google Scholar
- 26.Rosenthal S, Veloso M, Dey AK (2011) Is someone in this office available to help me? proactively seeking help from spatially-situated humans. Journal of Intelligent and Robotic Systems pp. 1–17 Google Scholar
- 27.Scaffidi C (2009) Topes: Enabling end-user programmers to validate and reformat data. Carnegie Mellon Technical Report CMU-ISR-09-105 Google Scholar