The Robotics Institute

RI | Centers | CFR | Seminar

Foundations of Robotics Seminar, November 1, 2006
Time and Place | Seminar Abstract | Speaker Biography | Presentation Slides | Speaker Appointments



From Path Planning to Motion Planning with Differential Constraints

Peng Cheng

 

Time and Place

Smith Hall 100
Refreshments 4:15 pm
Talk 4:30 pm

 

 

Abstract

 

The insufficiency of classical path planning for robotic systems subject to differential constraints in various applications, such as embedded system verification, micromanipulation, and air traffic control, motivates recent efforts to develop motion planning algorithms for systems with differential constraints (MPD).   Compared to the well understood path planning problems, MPD is still quite primitives in its problem formulation, complexity, decidability, algorithm design and analysis, and application exploration.  In this talk, I will present some of my recent results along this direction.  First, I will give a partial answer to the decidability question.  Based on an abstract formulation and an incremental decision algorithm, I present sufficient conditions for decidable MPD problems, which are applied on several non trivial MPD problems. For example, the existence of a trajectory for a DubinĄ¯s car with a rigid body between two configurations in a polygonal environment with a finite number of discontinuities in curvature is decidable.  Secondly, I will talk about the application of MPD in verification of controllers of embedded systems under uncertainties.  MPD algorithms are used to explore the uncertainty space looking for falsifying controls leading to controller failures, and resolution completeness is used to infer the controller safety when no falsifying controls are returned in finite time.  Lastly, I will talk about motion planning for a practical system, the roller racer.  Techniques from motion modeling, characterization, control, and planning are incorporated to design motions between any two positions and orientations with zero velocities using a single steering angle control. 

 

 

 

Speaker Bio

Peng Cheng received his Ph.D. in 2005 from department of computer science of University of Illinois at Urbana-Champaign under supervision of Dr. Steve LaValle, and now is working with Dr. Vijay Kumar as a Postdoctoral fellow in GRASP lab at University of Pennsylvania.  He is interested in studying motion planning, control, modeling, and characterization of various systems under differential constraints, with applications in verification, micromanipulation, multiple heterogeneous vehicle coverage, cable-driven robots, computational biology, and computer animation.    

 

 

Speaker Appointments

For appointments, please contact Howie Choset (choset@cs.cmu.edu)


The Robotics Institute is part of the School of Computer Science, Carnegie Mellon University.