Embedded Files
IMG_3391.MOVAutonomous obstacle avoidance for disaster recover navigation
Autonomous obstacle avoidance for disaster recover navigation
Zumi navigates a simulated damaged-building course using its front IR sensor to spot obstacles ahead. When it hits something, it stops, backs up, and turns 90 degrees in a random direction before moving forward again. Diagonal obstacles gave us trouble early on, since the IR beam would reflect away from the sensor and read as clear when it wasn't, so we recalibrated the threshold using real data from our course materials. We also tried a camera-based approach that compared brightness on each side to pick a smarter turn direction, but room lighting threw it off more than the obstacles did, so we scrapped it. The IR reverse-and-turn system ended up being the more reliable option. It clears multi-obstacle courses, diagonal walls included.
Title Here
Title Here
Description here
Title Here
Title Here
Description here
Title Here
Title Here
Description here
Please break up with headings. Idea:
Overview
Zumi navigates a simulated damaged-building course using its front IR sensor to spot obstacles ahead.
Basic Obstacle Response
When Zumi hits something, it stops, backs up, and turns 90 degrees in a random direction before moving forward again.
Challenges With Diagonal Obstacles
Diagonal obstacles gave us trouble early on, since the IR beam would reflect away from the sensor and read as clear when it wasn't. We recalibrated the threshold using real data from our course materials to fix this.
Camera-Based Approach (Abandoned)
We also tried a camera-based approach that compared brightness on each side to pick a smarter turn direction. However, room lighting threw it off more than the obstacles did, so we scrapped it.
Final Solution
The IR reverse-and-turn system ended up being the more reliable option. It clears multi-obstacle courses, diagonal walls included.
Page updated
Google Sites
Report abuse