Abstract
A longstanding barrier to deploying robots in the real world is the ongoing need to author robot behavior. Remote data collection–particularly crowdsourcing—is increasingly receiving interest. In this paper, we make the argument to scale robot programming to the crowd and present an initial investigation of the feasibility of this proposed method. Using an off-the-shelf visual programming interface, non-experts created simple robot programs for two typical robot tasks (navigation and pick-and-place). Each needed four subtasks with an increasing number of programming statements (if statement, while loop, variables) for successful completion of the programs. Initial findings of an online study (N = 279) indicate that non-experts, after minimal instruction, were able to create simple programs using an off-the-shelf visual programming interface. We discuss our findings and identify future avenues for this line of research.
Highlights
As robots move out of controlled industrial environments into the real world, a persistent challenge is the need to expand robot behavior to adapt and respond to real-world situations without constant expert supervision
It is important to recognize that our aim is to investigate if/how people can create simple robot programs, and this work is a first step towards nonexpert robot programming
Visual programming is a popular tool for non-expert programming, and it fits our robot programming through crowdsourcing use case as it allows for intuitive, non-expert robot programming without the need for extensive training
Summary
As robots move out of controlled industrial environments into the real world, a persistent challenge is the need to expand robot behavior to adapt and respond to real-world situations without constant expert supervision. Robot programming has primarily been a task for engineers that required a high level of mathematical and programming knowledge. This has limited the amount of programs created and, in turn, versatility. While short-horizon skill learning, such as pushing and grasping objects, has typically been taught through small-scale learning from demonstration (LfD) (Argall et al, 2009) or teleoperation-like control, higher-level behavior (the order of actions to accomplish a task successfully) or additional rules (move to the room, but if the door is closed open the door first) could be scaled to the crowd and directly provided in the form of simple robot programs. In a pick-and-place task, if a robot’s task changes from picking up any item from a tray to picking up only pink items, a nonexpert crowdworker anywhere on earth, at any time of the day, can provide a rule that specifies in Frontiers in Robotics and AI | www.frontiersin.org van Waveren et al
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
