Implement curb detection

[5yler]

New York Maker Faire track might not have solid barriers all around. Possible solutions:

• downward-tilted LIDAR
• ZED stereo camera

The ZED could be integrated with depthimage_to_laserscan or pointcloud_to_laserscan. The latter is less efficient but seems to allow conversion between frames, which would allow greater flexibility in sensor mounting.

Tasks

• [x] update gigatron.urdf.xacro to include:
  • [x] ZED
  • [x] upside-down XV LIDAR on front of car
  • [x] IMU mount location
• [ ] update gigatron.launch to apply appropriate laser angle filters
• [ ] rewrite pointcloud_to_scan.launch to take max, min height parameters, output scan topic name, and output frame_id

[5yler]

Depth cloud from ZED camera

[5yler]

Two laser scans generated from the ZED point cloud:

• red: sampled from points with z between 0 and 0.5 m
• orange: sampled from points with z between 0.5 and 2m

[5yler]

XV LIDAR is now mounted under the front of Gigatron's body. Need to update launch files and URDF model accordingly.

[5yler]

Bottom-mounted RPLIDAR performed well on the track at World Maker Faire 2016 in NYC. Videos from the track: