Post on 20-Feb-2016
description
Autonomous Soil Investigator
What Is the ASI?
• Designed to complete the 2013
IEEE student robotics challenge
• Collects "soil" samples from a simulated forest environment
The ASI Solution
• Localizationo Hypobotso Sensor Input
• Traversalo Pathfindingo Wheels
• Collectiono Armao OpenCV / Inverse Kinematics
Localization
Position and Orientation Awareness
Localization sensors:
Four eye modules, each with• Infrared rangefinder• Ultrasonic rangefinder• Sweeping Servo Motor
How was Localization implemented
Particle Filter:• Uses a cloud of discrete
hypotheses (Hypobots) of the robot's position
• Cloud mimics robot's intended motions
• Each time a measurement is performed, each hypothesis is weighted
Localization lessons learned:
• Work with the robot's perception, rather than your own
• Robustness is more important than efficiency
Pathfinder
New Data:• Localization• IMU• Pucks
Static Data:• Graph Data
Planning and Routing
PathfindersProbabilistic
pathfinding
Planned Pathingvs.
Probabilistic pathfinding
• Slow to navigate in X, Y, Theta space• Cannot find tricky solutions• Paths are often not optimal• Non-Voronoi solutions
• Quick to solutions
• Location specific conditions
• Trick solutions
• Custom GUI
Planned Pathing
Collection
Target Acquisition
OpenCV - Computer Vision tool library
Used to precisely locate samples
Inverse Kinematics
Hardware -- Panda Board
• Featureso Dual Core, 1.2 GHz ARM Processoro Ubuntu 12.04 Linux Nativeo USB Host Controller
• Purpose: High Level Computingo Localization Algorithmso Pathfinding Algorithmso Computer Vision
Hardware -- Microcontroller• Features
o Rapid prototypingo Common Tools
• Purpose: Lower Deck Computingo Ackerman geometryo PWM for servo & motor controlo ADC for infrared o Sonar sensor interfacing
Architecture
Project Structure and Optimization
Collaboration Optimization
Operation Methodology
Methodology
MESSENGER
CONTROL
MODE
BASE
Panda Board
Python
Arduino
C/C++
• Operations per mode
• Binds modules
• Messenger communication central
• Stages
Methodology (continued)
• Polymorphism
• Transitions
• Tailored behaviors
Init
Abstract Mode
CollectLocalize Pathfinding . . .
Collaboration
• Github
o Distributed workflow
o Quality
o Recovery
TEAM IMAGE