Hello roboticists, I will be presenting the talk for my preliminary exam tomorrow, Tuesday, April 18th at 12 PM. Everyone is welcome to attend in person or virtually, per details below. *Title:* User-Friendly Robot Planning for Scientific Data Collection *Date & Time:* Tuesday, April 18th at 12:00PM *Location:* Rogers 226 *Zoom Link:* https://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Foregonstate.zoom.us%2Fj%2F97231249962%3Fpwd%3DUktjNTdYcFZCKzZlNEswREVCRkZjQT09&data=05%7C01%7Crobotics-seminar%40engr.orst.edu%7Ccb8655ed5f4e418f692c08db3f71b209%7Cce6d05e13c5e4d6287a84c4a2713c113%7C0%7C0%7C638173529767580409%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=NsGhvodsLXKHZYXLSyAx8Ky8UIoUUYUED3Lz1SEMuYs%3D&reserved=0 Password: 673814 *Abstract:* In this thesis, we propose methods to improve the interaction between decision making algorithms for robotic systems and the human users. Why do we care if a scientific data collection robot is user-friendly? The traditional approach to human robot interaction looks at robots that physically inhabit the same space as humans. However, robots used for scientific data collection tasks typically do not inhabit the same space as humans, such as planetary exploration rovers. These robots are still commanded by humans, and their interaction must be carefully considered as there is limited time that humans and the robot can interact. Similar to the planetary exploration example, many other scientific data collection robots, such as ocean monitoring, require complex high-level decision making algorithms to maximize their performance. Before a robotic system can make decisions, a user must define the constraints and objectives of the robot. Additionally, users prefer robotic systems that make understandable decisions. Defining the constraints, objectives, and understanding the output from these high-level decision making algorithms is a challenging problem. In order to advance user-friendly robot planners, we propose three contributions. The first contribution improves constraint definition by mapping language instructions into topological constraints a robot planner can follow. The second contribution allows objective definition by learning tradeoffs between potentially competing objectives by asking users to select, rank, or rate a set of paths. We propose a method that selects different query types to show to the user that allows them to characterize their own objectives without having to explicitly define the function. The third contribution enables understanding of decision making algorithms by adding explainability to their choices using insights from the social sciences. In this third contribution, we additionally propose combining explainability and user preferences to allow the user to elucidate their preferences using an interactive user preference method. These contributions allow users to efficiently provide input to the robotic system to define constraints and objectives. Together, these contributions combine to enable both efficient definition of constraints and objectives, as well as improved interpretability of robotic decision making algorithms for scientific data collection. *Advisor:* Dr. Geoff Hollinger *Committee Members:* Dr. Julie Adams, Dr. Alan Fern, Dr. Maria Kavanaugh, Dr. Stefan Lee Thank you! Ian