Presentation is loading. Please wait.

Presentation is loading. Please wait.

CS 7455 Term Project Robot control by Machine learning

Similar presentations


Presentation on theme: "CS 7455 Term Project Robot control by Machine learning"— Presentation transcript:

1 CS 7455 Term Project Robot control by Machine learning
William Hawkes Mobile Application Development

2 Overview Project Description Project Procedure/Methodology
Objective Results Project Procedure/Methodology Robot construction Android App – Raspberry Pi 3 Algorithms used for Computer Vision Key Findings/Results Results obtained with robot construction Results obtained using Computer Vision Conclusion

3 Project Description Objective Results
To use an Android phone to relay commands and control a robots movements To develop an Android app that takes camera images, uses image processing and completes the following tasks Follow line drawn on paper Turn to follow motion. Results Robot construction is complete Fairley successful results with Function1 and 2

4 Procedure/Methodology
Robot Construction Two designs $12 kit from EBay (Figure 1) Great starting point Not very sturdy or accurate. Used Arduino with Bluetooth Serial – far too much delay Custom design using kit as reference (Figure 2) Much sturdier construction Uses higher quality motors Used Raspberry Pi 3 with built-in WiFi and Bluetooth – much more powerful and responsive

5 Procedure/Methodology
Android App – Raspberry Pi 3 Android application using OpenCV library and Bluetooth communication Raspberry Pi using control program written in c Communicates with Android through Bluetooth serial port

6 Procedure/Methodology
Algorithms used for Computer Vision Simple command interface with Raspberry Pi 3 CMD:FWD CMD:REV CMD:LT CMD:RT CMD:STOP CMD:REBOOT CMD:SHUTDOWN CMD:RL:XX CMD:RR:XX Display Image

7 Procedure/Methodology
F1 Following a line

8 Procedure/Methodology
F2 Follow motion

9 Procedure/Methodology
F2 Follow motion (continued)

10 Procedure/Methodology
Figure 1) EBay kit using Arduino

11 Procedure/Methodology
Figure 2) Custom design using Raspberry Pi3

12 Procedure/Methodology
Controller board – prototype (left) finished board (right)

13 Procedure/Methodology
Custom Buttons: Create image, save in drawable-[display size] Create xml file below in drawable (forward_button.xml) Add button to view with background set to your image Android Application – View of motion control (F2)

14 Key Findings/Results Steering method used in this project did not allow adequate response time to follow a line with much deviation. Auto focus of camera an issue, especially with motion follow.

15 Conclusion Overall happy with the results of this project
Learned a lot about using OpenCV Furthered my knowledge of Android development First experience using Raspberry Pi 3 and hardware interfacing with it Improvements Different steering mechanism. Improved algorithm to detect motion. Try image recognition. Error detection. Try using midpoint function for motion detection.

16 Questions and Discussion


Download ppt "CS 7455 Term Project Robot control by Machine learning"

Similar presentations


Ads by Google