Learning outcomes
- Explain VIO without using estimator jargon
- Know why cameras and IMUs are used together
- Identify the benchmark output as a timestamped trajectory
The basic task
Visual-inertial odometry estimates how a moving platform changes position and orientation over time.
The camera sees the scene. The IMU measures fast rotation and acceleration. VIO combines both so the estimate can keep moving between camera frames.
In this benchmark, the public output is a trajectory file: one pose row per timestamp.
What the benchmark does not ask for
The public benchmark is not asking each system to build a global map or revise old poses with future information.
The primary trajectory is the odometry output available at the current time step.
Checklist
- Can I point to the camera stream?
- Can I point to the IMU stream?
- Can I point to the pose trajectory the system writes?
Practice task
Open the Results page and find the pose artifact column. Do not interpret any TBD value as a numeric score.