Skip to content

Test Plan

We have broken down our system tests into Fall and Spring Validation Experiments. They are outlined in detail below, as well as in these documents: FVE, SVE.

In addition, we have made a complete test plan for the Spring Semester with intermediate goals and verification. The document can be seen here.

[Completed] Spring Validation Experiment – Apr 26/ May 3

Location: Newell-Simon Hall, B floor

Test setup

  • Size: minimum 6m x 3m space
  • Floor surface: Smooth tiling
  • Static obstacle: 0.15m x 0.15m x 0.2m or larger
  • Road: 2 lanes
  • Lane width: 0.60m*

Initial procedure for all tests

  1. Initialize robot in known environment, on the right lane of the two-lane road.
  2. Send command velocity : straight forward with constant velocity of 3 m/s** while staying in the lane.
  3. When front of robot is ~1.5 m away from predetermined obstacle position, the obstacle will be introduced.

Demonstration procedure 1

  1. Initial procedure (see above)
  2. The robot will see the obstacle and execute an emergency brake.
  3. Observe the resulting trajectory of the robot.
    1. Robot will crash into obstacle in its attempt to come to a stop.

Demonstration procedure 2

  1. Initial procedure (see above).
  2. When the obstacle is introduced, the human operator will be given manual control of the vehicle via a joystick. The operator will attempt to avoid the obstacle.
  3. Observe the resulting trajectory of the robot.
  4. Repeat steps 1-3 three times.

Test procedure

  1. Initial procedure (see above).
  2. The robot will see the obstacle and execute a maneuver around it, while staying within the limits of the adjacent lane.
  3. Observe the resulting trajectory of the robot.

Success criteria

  • Robot detects obstacle on current path (show on laptop).
  • Robot avoids contact with obstacle
  • Robot resumes its original trajectory (straight forward) along the lane.
  • Robot stays within road markings throughout the test.

* US Highway typical lane width: 12 feet (3.7m); at 1/10 scale, lane widths will be about 0.37m. However with our crash bumper, the footprint of the robot is enlarged by 1.5 times, requiring a lane width of >0.55m

** Freeway speeds are usually > 60mph ~= 2.7 m/s at 1/10 scale. Moose tests are typically conducted around 40mph ~= 1.8m/s

[Completed] Fall Validation Experiments – December 1st/8th

 

fve

Location: Newell-Simon Hall, B floor

Test setup

  • Size: minimum 6m x 4m space
  • Floor surface: smooth tiles or concrete
  • Static obstacles: 0.15m x 0.15m x 0.2m or larger

Test procedure

Test 1: Localization Accuracy Test

  1. Mark 3 points in test environment with tape
  2. Initialize the robot at any one of the points
  3. Tele-operate the robot to any one of the other marked points
  4. Obtain the position estimate of the robot
  5. Compare the position estimate with the true coordinates of the marked point

Test 2: Navigation and Obstacle Avoidance Test

  1. Place one or more obstacles in the environment, which are not part of the known map
  2. Initialize robot with known map
  3. Send goal point via GUI
  4. Robot will travel from start point to goal point at a commanded velocity of 1 m/s
  5. Measure the distance of the robot from the goal point
  6. Repeat steps 1~4, two more times

Test 3: Simulation model for drifting

  1. Input commands and initial conditions to Simulink model
  2. Observe animation of the simulated motion of the robot
  3. Input an equivalent set of commands (steering angle, throttle) to robot
  4. Observe motion of robot in real world

Success criteria

Test 1: Localization Accuracy Test

  • Robot outputs an estimate of its position, relative to the map
  • Position: Robot localizes itself within 0.15m of each marked point

Test 2: Navigation and Obstacle Avoidance Test

  • Robot plans an initial trajectory to its destination
  • Robot sends updates of its planned trajectory and map to GUI
  • Robot avoids known and unknown static obstacles
  • Robot reaches planned destination coordinates within 0.15m

Test 3: Simulation model for drifting

  • Model is able to drift in a predictable fashion in the simulation
  • Robot is able to achieve drifting behaviour in real world test