From AIRWiki

Jump to: navigation, search
Search-based Differential Drive Planner
Short Description: A Search-Based Trajectory Planner for differential drive vehicles in ROS Context
Coordinator: MatteoMatteucci (matteo.matteucci@polimi.it)
Tutor: MatteoMatteucci (matteo.matteucci@polimi.it)
Students: VitoRessa (vito.ressa@mail.polimi.it)
Research Area: Robotics
Research Topic: Robot development
Start: 11/2014
Status: Active
Level: Ms
Type: Thesis


Search-based Differential Drive Planner


The aim of the project is to develop a global path planner for a robotic wheelchair (viewed as a Differential Drive Robot) to provide feasible (i.e. dynamically consistent) trajectories to the robot to be followed.

Problem approach

The planner will be search-based with the usage of 3-variables(standard SBPL) and 5-variables (new custom development) environments to ensure dynamical feasibility to the plans generated.

Motion primitives will be generated solving various optimal cotrol problems highly customizable to have different set of primitives for different kind of differential drive robots. There are various methods for motion primitives generation (multi-resolution, "intelligent" reduction, ... ).

The planner will be written as a move_base plugin to have a robust and flexible environment in which the new approach could be tried, understood and compared.

Technologies used

The planner is search-based, using SBPL as a base library and is developed as a move_base (i.e. a ROS package in navigation stack) plugin. Tools for primitives generation use GPOPS as a optimal control problems solver.

The code for the planner and the benchmark tools is written in C++; Primitive generation tools and utilities for visualization are provided as Matlab functions.