Download presentation
Presentation is loading. Please wait.
Published byGeoffroy Petit Modified over 6 years ago
1
Aaron Swenson Samuel Farnsworth Derek Stewart Craig Call
2
Agenda Previous Action Items State of project mechanical vision
AI/architecture controls Items from FSD… is it up to spec? Demonstration Conclusion
3
Previous Action Items Mechanical
Rework component placement, including replacing breadboard with vectorboard Update power connectors to make the system more stable AI/Architecture Complete AI library and play strategies Implement Kalman filter Implement path tracking and prediction continued...
4
Previous Action Items continued...
Controls Implement path planning and basic motor controls Thoroughly test current control architecture Vision Remove fisheye effect from video Track shapes in addition to color - found not necessary
5
Overview… hasn’t changed
6
Mechanical System - The Good
Heat shrink bundles Small vector board Robot casing
7
Mechanical System - The Bad
Floating ground Burned out encoders
8
Is it up to spec? # Metric Min/Max Value Ideal Value Actual Value 1
Top Speed 12 in/sec 16 in/sec 34 in/sec 2 Total Power Consumption < 50W < 40W ~4W idle ~6W running 3 Radius of Communication 10 meters 20 meters ~35 meters
9
Vision - The Good Fixed Camera Skew Before After
10
VISION - The Good Added an easy-to-use controller
Easily pick what we track Switch instantly from Home to Away Easily adjust for camera/lighting changes
11
VISION - The Latency Image segmentation Looks at small area around
objects to speed up tracking Slow Still about a second of latency
12
Computer Architecture - The Good
ROS nodes communicate beautifully ROS launch file functional Libraries easily accessible on either the desktop or odroid
13
Computer Architecture - The Bad
ROS took longer to complete than expected Parallelized computing was more difficult to implement in ROS Not necessarily immediately expandable to a two robot system
14
Artificial Intelligence
Processed Vision Information Implemented Rush Goal Oriented toward the goal
15
Artificial Stupidity Only did one play/strategy
Latency affects play/strategy selection Class decision delay
16
Control System - The Good
Motion and direction Architecture Ros nodes for each level Testing Easy to change and update Quick to test
17
Control System - The Bad
The matrices (M- and rotation) Orientation (theta?) Delays ROS Encoders Vision latency (1½ sec delay!)
18
Difficulties Faced Floating Ground Vision Lag Control Matrices
Team Communication
19
All’s Well That Ends Well
Group dynamic Suggestions for next semester Camera Jersey Other hardware and equipment Problems/Issues Page Commonly encountered issues and fixes
20
Questions? Comments?
21
Questions? Comments?
22
Questions? Comments?
23
OLD SLIDE SHOW FOR REFERENCE AFTER THIS POINT
24
Overview
25
Previous Action Items Mechanical: Complete robot design
Build second robot Vision: Robot and ball detection Kalman Filter implemented Controls: Implement PID control Speed and distance controlled commands (move to any point) AI/Architecture: ROS setup and communicating between systems Implement 1 offensive and 1 defensive play
26
Milestones Acheived Mechanical: Complete robot design
Build second robot Vision: Robot and ball detection Kalman Filter implemented Controls: Implement PID control Speed and distance control commands (move to any point) AI/Architecture: ROS setup and communicating between systems Implement 1 offensive and 1 defensive play
27
Difficulties Faced Build second robot:
Still working on first robot, parts have not yet arrived. Improve design before replication Kalman Filter implemented: Camera specifications kept changing Distance control commands : PID Controller took longer to complete than anticipated Implement 1 offensive and 1 defensive play: High dependency on Vision
28
Schedule
29
Schedule continued...
30
Schedule continued...
31
Schedule continued... The following tasks do not have a set start date or duration: Maintain code base (Sam) Optimization (Sam) Debug algorithms (Sam) Refine Vision (Derek) Perform routine maintenance (Aaron)
32
Program Status Mechanical Reworked robot frame
Added wiring for encoder feedback Easy battery/power cord switching Vision Color identification algorithm working Tracking ball and robot position (cm) and orientation (d Controls Implemented velocity controller and trajectory tracker Designed M-matrix (rotational matrix) logic AI/Architecture ROS up and running on robot and base station Refactored code base for easier use AI library code base started
33
Program Status - New Robot Frame
Bottom Layer Middle Layer Top Layer
34
Action Items Mechanical
Rework component placement, including replacing breadboard with vectorboard Update power connectors to make the system more stable Vision Remove fisheye effect from video Track shapes in addition to color continued...
35
Action Items continued...
Controls Implement path planning and basic motor controls Thoroughly test current control architecture AI/Architecture Complete AI library and play strategies Implement Kalman filter Implement path tracking and prediction
36
Questions? Comments?
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.