Video Showcase This page was initiated by Claudia Esteves
Motion Planning for Cooperating Virtual Characters
Title: Motion Planning for Cooperating Virtual Characters
Authors: C. Esteves, G. Arechavaleta, J. Pettré and J-P. Laumound
Institution(s): LAAS-CNRS - Gepetto research group
Abstract: This work presents an approach to automatically compute animations for antropomorphic virtual characters that cooperate to move objects in cluttered environments. The main challenge is to deal with 3D collision avoidance while preserving the believability of the character's behaviours. To accomplish the coordinated task, a geometric and kinematic decoupling of the system is proposed. Different techniques such as probabilistic path planning, locomotion controllers, inverse kinematics and path planning are merged into a centralized planner.
Keywords: autonomous characters, behavior modeling, motion control, motion planning
Interactive Motion Correction and Object Manipulation
Title: Interactive Motion Correction and Object Manipulation
Authors: Ari Shapiro, Marcelo Kallmann, and Petros Faloutsos
Institution(s): UC Merced and UCLA
Full Video: mp4 QuickTime 40MB
Sample Videos: kallmann-movingring.mov kallmann-umbrella.mov kallmann-manip.mov
Abstract: Editing recorded motions to make them suitable for different sets of environmental constraints is a general and difficult open problem. In this paper we solve a significant part of this problem by modifying full-body motions with an interactive randomized motion planner. Our method is able to synthesize collision-free motions for specified linkages of multiple animated characters in synchrony with the characters full-body motions. The proposed method runs at interactive speed for dynamic environments of realistic complexity. We demonstrate the effectiveness of our interactive motion editing approach with two important applications: (a) motion correction (to remove collisions) and (b) synthesis of realistic object manipulation sequences on top of locomotion.
Motion Planning of Multiple Agents in Virtual Environments
Title: Motion Planning of Multiple Agents in Virtual Environments
Authors: Yi Li and Kamal Gupta
Institution: Simon Fraser University
Lab: Robotic Algorithms and Motion Planning Lab (RAMP)
Videos: MP4 Video One, MP4 Video Two, MP4 Video Three
Abstract: We are interested in real-time motion planning of multiple agents in two dimensional virtual environments (given as binary occupancy grids). In the first video, we present an adaptive multi-resolution continuum model for motion planning of multiple agents. Compared to the basic continuum model, our approach allows each agent to have its own start position and goal position, while maintains the advantages of the basic continuum model such as unified global planning and collision avoidance. Our approach intrinsically prefers open areas (if desired). This is a desirable trait since otherwise deadlocks may occur. However, if a narrow passage is the only choice and agents have to pass through it, we could augment the continuum model with coordination graphs for deadlock avoidance as shown in the second video. The third and final video shows a novel approach for real-time motion planning of multiple formations in virtual environments. This algorithm is based on the continuum model and our flexible virtual structure approach for formation control.
Keywords: formation, motion coordination, motion planning, multiple agents, real-time, virtual environments
Large-Scale Real-time Modeling of Multiple Human Agents
Title: Large-Scale Real-time Modeling of Multiple Human Agents
Authors:Avneesh Sud, Russell Gayle, Stephen Guy, Erik Andersen, Ming Lin and Dinesh Manocha
Institution(s):University of North Carolina at Chapel Hill - Gamma Team
High-resolution video:http://gamma.cs.unc.edu/crowd/largecrowd/
Abstract: We present a real-time algorithm for simulating heterogeneous crowds with potentially distinct individual behavior characteristics and goals. Our approach combines global motion planning with a generalized pedestrian dynamics model to synthesize motion for numerous groups of people in complex dynamic environments. Inspired by self-organization phenomena in crowds, we introduce "Pedestrian Levels of Detail" to accelerate large-scale simulation of individual agents, while preserving natural collective behaviors observed in real crowds. We highlight the performance of our algorithm on heterogeneous crowds in highly populated urban scenes.
Keywords: autonomous characters, motion planning, virtual crowds
Task-consistent motion generation in dynamic environments
Title: Task-consistent motion generation in dynamic environments
Authors: Yuandong Yang and Oliver Brock
Institution(s): UMass Amherst
Abstract: We present the Elastic Roadmap motion generation method that enables autonomous mobile manipulators to the perform tasks robustly in dynamic environments. The videos show how 1) the UMan mobile manipulator performs a line-following task among unpredictably-moving obatcles; 2) a stationary manipulator reaches an object in a moving truss, while maintaining end-effector orientation; 3) the UMan mobile manipulator follows unpredictably-moving object and recovers from failures.
Keywords: autonomous mobile manipulation, task-consistent, motion control, motion planning
Balancing Exploration and Exploitation in Motion Planning
Title: Balancing Exploration and Exploitation in Motion Planning
Authors: Markus Rickert, Oliver Brock, Alois Knoll
Institution(s): Technische Universität München, UMass Amherst
Abstract: Computationally efficient motion planning avoids exhaustive exploration of high-dimensional configuration spaces by leveraging the structure present in real-world planning problems. We argue that this can be accomplished most effectively by carefully balancing exploration and exploitation. Exploration seeks to understand configuration space, irrespective of the planning problem, while exploitation acts to solve the problem, given the available information obtained by exploration. We present an Exploring/Exploiting Tree (EET) planner that balances its exploration and exploitation behavior. The planner acquires workspace information and subsequently uses this information for exploitation in configuration space. If exploitation fails in difficult regions, the planner gradually shifts to its behavior towards exploration. We present experimental results to demonstrate that adaptive balancing of exploration and exploitation leads to significant performance improvements, compared to other state-of-the-art sampling-based planners.
Keywords: adaptive, exploration, exploitation, workspace information, motion planning
Planning 3D Collision-Free Dynamic Humanoid Motion through Iterative Reshaping
Title: Planning 3D Collision-Free Dynamic Humanoid Motion through Iterative Reshaping
Authors: E. Yoshida, C. Esteves, I. Belosouv, J-P. Laumond, T. Sakaguchi and K. Yokoi
Institution(s): AIST-CNRS Joint Japanese-French Robotics Laboratory
Abstract: An iterative planning method is proposed to generate 3D collision-free motions that take account of dynamics of complex robots. The planner consists of two stages, sampling-base path planning and dynamic motion generation. We introduce an spatial reshaping method when collisions are found at the end of an iteration. We demonstrate the effectiveness of the proposed method with a humanoid robot executing dynamic manipulation and locomotion at the same time.
Keywords: motion planning, collision-avoidance, dynamics, humanoids.
Next Video Title
Just copy from above.