© sebastian thrun, CMU, C Statistical Techniques In Robotics Sebastian Thrun and Geoffrey Gordon Carnegie Mellon University

© sebastian thrun, CMU, notes  Pointer to Larry’s material

© sebastian thrun, CMU, Administrative Information  Sebastian Thrun  Geoffrey  Web:  list:  Time:Mon/Wed, 10:30-11:50am  Location:NSH 3302  TA:n/a  Appointments: send !

© sebastian thrun, CMU, 20004

5 Goals  Enable you to program robots and embedded systems in a robust fashion  Enable you to understand the intrinsic assumptions in your robot software  Enable you to pursue original research in probabilistic robotics  Sway you into joining a young and fascinating research field: probabilistic robotics

© sebastian thrun, CMU, What this course is not  Intro to robotics  Little work  Low on math

© sebastian thrun, CMU, Course Schedule Localization Sept 4-16 Mapping Sept 30-Oct 16 Decision Making Oct Multi-Agent Nov 4-11 Advanced Perception Nov 13-25

© sebastian thrun, CMU, What You Should Do  Think  Think differently  Be critical  Come up with Original Research

© sebastian thrun, CMU, What Is A Good Project  Mine Mapping  Multi-Agent Control

© sebastian thrun, CMU, Requirements  In teams of three: Warm-up project (mobile robot localization) Written assignment(s) Research Project  Class Presence: all but two sessions (send me )  Quizzes (all but at most two)  No exams

© sebastian thrun, CMU, Your next tasks  Check out Web site Read assigned paper Download map+sensor data and program robot localization algorithm  Send Sebastian mail with your name and names of team mates (for warm-up project)  Come to class on Sept 9 th (10:30am-11:50am)

© sebastian thrun, CMU,

© sebastian thrun, CMU,

© sebastian thrun, CMU,

© sebastian thrun, CMU,

© sebastian thrun, CMU, Five Sources of Uncertainty Environment Dynamics Random Action Effects Sensor Limitations Inaccurate Models Approximate Computation

© sebastian thrun, CMU, Trends in Robotics Reactive Paradigm (mid-80’s) no models relies heavily on good sensing Probabilistic Robotics (since mid-90’s) seamless integration of models and sensing inaccurate models, inaccurate sensors Hybrids (since 90’s) model-based at higher levels reactive at lower levels Classical Robotics (mid-70’s) exact models no sensing necessary

© sebastian thrun, CMU,

© sebastian thrun, CMU, Rhino

© sebastian thrun, CMU, Minerva

© sebastian thrun, CMU, The CMU/Pitt Nursebot Initiative

© sebastian thrun, CMU, People Detection Mike Montemerlo

© sebastian thrun, CMU, Learning Models of People Maren Bennewitz

© sebastian thrun, CMU, D Mapping Result With: Christian Martin

© sebastian thrun, CMU, Multi-Robot Exploration

© sebastian thrun, CMU, Mine Mapping (brand new)

© sebastian thrun, CMU, What are interesting problems?  Mapping, automatic, manual, guided?  Probabilistic localization, landmarks?, odometer!,  Route planning, collision avoidance  Mine Mapping?

© sebastian thrun, CMU, How can we solve them?

© sebastian thrun, CMU,

© sebastian thrun, CMU, Where Am I/?

© sebastian thrun, CMU, Nature of Sensor Data: Uncertainty Odometry Data Range Data

© sebastian thrun, CMU,

© sebastian thrun, CMU, Warm-Up Assignment: Localization, Due Sept 23

© sebastian thrun, CMU, Warm-Up Assignment: Localization

© sebastian thrun, CMU, Warm-Up Assignment: Localization