Robotics & ROS test (Intermediate)

Sample questions

5 sample questions from the Robotics & ROS test, with the answers. Samples are taken out of the live test, so you will not meet these when you sit it.

1. What is the expected outcome of sensor fusion in a multi-robot system when using complementary sensor data, such as LiDAR and IMU?

   1. Sensor fusion will not affect the accuracy or noise of the estimate.
   2. The fused estimate will be more accurate and less noisy due to complementary sensor data.
   3. The fused estimate will be identical to the data from the most accurate sensor.
   4. Sensor fusion may introduce more noise due to the combination of multiple data sources.

   Answer: B. The fused estimate will be more accurate and less noisy due to complementary sensor data.

   Sensor fusion in multi-robot systems combines data from multiple sensors, such as LiDAR, IMU, and GPS, to improve accuracy and reduce noise. When sensor data is complementary (e.g., LiDAR provides accurate distance measurements while IMU provides angular data), the fused estimate becomes more reliable. This is because each sensor has its own strengths and weaknesses, and by combining them, the system can mitigate errors and produce a more robust estimate.

2. Which statement correctly describes the behavior of the Kalman filter's covariance matrix P during the prediction step in the context of sensor noise modeling?

   1. The Kalman filter's covariance matrix P shrinks during the prediction step due to process noise.
   2. The Kalman filter's covariance matrix P grows during the prediction step due to process noise.
   3. The Kalman filter's covariance matrix P is unaffected by process noise during the prediction step.
   4. The Kalman filter's covariance matrix P grows during the update step due to process noise.

   Answer: B. The Kalman filter's covariance matrix P grows during the prediction step due to process noise.

   The Kalman filter's prediction step accounts for process noise by increasing the covariance matrix P, which represents the uncertainty in the system's state. This is a fundamental part of how the filter models sensor noise and uncertainty. Other options incorrectly describe the behavior of the Kalman filter or its components.

3. In which scenario is particle filter resampling preferred over the Kalman filter, and why?

   1. Particle filter resampling is preferred over the Kalman filter in all scenarios due to its superior accuracy and versatility.
   2. Particle filter resampling is preferred over the Kalman filter when dealing with non-Gaussian and multimodal distributions because the Kalman filter assumes Gaussian distributions and cannot effectively represent complex, non-linear, or multi-peaked probability distributions.
   3. Particle filter resampling is preferred over the Kalman filter only when the system is linear and the noise is Gaussian.
   4. Particle filter resampling is preferred over the Kalman filter when the system is deterministic and has no noise.

   Answer: B. Particle filter resampling is preferred over the Kalman filter when dealing with non-Gaussian and multimodal distributions because the Kalman filter assumes Gaussian distributions and cannot effectively represent complex, non-linear, or multi-peaked probability distributions.

   Particle filter resampling is a technique used in particle filters to address the problem of sample degeneracy, where a few particles dominate the distribution. It is particularly useful in scenarios where the belief distribution is non-Gaussian and multimodal, as particle filters can capture these complex distributions. On the other hand, the Kalman filter is optimal for linear systems with Gaussian noise but may fail to accurately represent non-Gaussian or multimodal distributions. Therefore, particle filter resampling is preferred in such cases, while the Kalman filter is better suited for linear and Gaussian scenarios.

4. In localisation, what does the likelihood calculation represent?

   1. The likelihood is proportional to the probability of the sensor measurement given the robot's pose.
   2. The likelihood is calculated based on the map's resolution.
   3. The likelihood depends on the number of particles in the particle filter.
   4. The likelihood is determined by the robot's prior position estimate.

   Answer: A. The likelihood is proportional to the probability of the sensor measurement given the robot's pose.

   In localisation, the likelihood calculation determines how probable a sensor measurement is given a specific robot pose. This is a critical step in methods like particle filters, where each particle's weight is updated based on the likelihood of the measurement. The likelihood is not about the robot's prior position, the number of particles, or the map's resolution, but about the sensor data's probability given the current pose estimate.

5. Which of the following is the most effective method to ensure sensor data from multiple sources (e.g., wheel odometry, IMU, GPS) is synchronized before being fused in a Kalman filter?

   1. Use a Kalman filter to estimate the correct timestamps during fusion.
   2. Use a time synchronisation protocol like PTZ (Precision Time Protocol) to align timestamps across sensors before fusion.
   3. Use a fixed offset between sensor timestamps for alignment.
   4. Ignore timestamps and assume they are already aligned.

   Answer: B. Use a time synchronisation protocol like PTZ (Precision Time Protocol) to align timestamps across sensors before fusion.

   Synchronizing sensor data is crucial for accurate sensor fusion. Using a time synchronisation protocol like PTZ ensures that all sensors share a common time reference, which allows for accurate alignment of data. This is especially important in systems like robot_localization, where data from multiple sensors (e.g., wheel odometry, IMU, GPS) must be merged. Other options are either incorrect or not directly related to solving the synchronization issue.