AI

Autonomous Challenge – AI

*Output:**
Vector determining direction + time + speed

Initialization:
No info about track just initial location.

Running:
*Run
- list item
- Line detection on camera image try to detect track limits in front of robot.
- If lines in sight then:
- Determine direction of movement of robot: middle of distance between 2 lines
- If no lines in sight assume same directions as before (since lines may be hashed)
- Preprocess lidar data over the 10 ft width of the track up until a length ahead of x meters (3-4m) and determine what side of the track is more favorable.
- Decide on movement both parallel to lines and coherent with current global direction
- Run circle algorithm both on cam and lidar:
*No collision move ahead and take new data (cam )
*Collision:
- Try moving circle left and right and reducing it till robot can move. Then repeat
- Update global direction for future reference

Concerns:
- Pot holes? CAM
- Camera capabilities? In terms of line detection: we need it to determine curves
- Need for map?? NO
- How do we know if things are done?

Navigation challenge AI

*Output:** vector indicating direction of movement of the robot + time + speed
Speed and time are dependent on size of circle and how close to target.
Vector’s origin could be either where we estimate the robot is now or where the robot was at time where data was collected. This is dependent on how fast the data processing is done.

Assumptions: use global map

Initialization:
*Initial map: start location + location of goals.
*Assume no obstacles and choose goal returned by A*

Running:
- Run A*
- Determine optimal direction
- Data Taken
- Project circle in direction given by A*
- Data processing
- 2 cases:
+No collisions -> determine speed and time -> send commands
+Collisions :
- Try some shifting left and right. Then smaller circle sizes and shifts. If doesn’t work pop back and update global map and call A*
- Repeat until no collision.

Marking obstacles:
- If circle collides even after small movement left or right then increment “gray color” of underlying region of squares. Before making circle smaller.

Concerns:
- Parameters
- If can use lidar for updating global map
- Camera detects what?

Unless otherwise stated, the content of this page is licensed under Creative Commons Attribution-ShareAlike 3.0 License