COM2009 L12 SLAM

What is SLAM?

Simultaneous Localization and Mapping, a fundamental problem in robotics.

What is state tracking?

Robust tracking of the current pose relative to the environment.

What are the two sources of information for state tracking?

Model (predictions) and sensor (updates using external cues)

What is the core problem for all autonomous navigation methods?

Accurately tracking the current pose relative to the environment.

What is the Kalman Filter?

An effective, recursive filter that estimates the internal state of a linear dynamic system from series noisy measurements.

What does the Kalman Filter allow?

Sensory updates to predicted state.

What is the Kalman Gain?

the influence that sensory inputs have on the overall output determined by its reliability (reliable sensory inputs will have a higher influence on the output)

What is the purpose of building a map of the local environment?

To estimate the state of the object in the world.

What are the steps involved in building a map?

1. Where am I?

2. Measure distance to object using LiDAR & updates map model

3. Robot moves and we make a predictions of what the world model might look like in the new pose.

4. More LiDAR measurement are obtained.

5. Sensory data used to update the map.

What is the purpose of predicting state from a model?

To calculate the predicted outcome using the input to robot and the robot model

What does the robot model consist of?

Previous state estimate

Impact of the input signal

System noise(Normal Distribution)

What happens when 2 independent normal distribution is summed?

The variance increases. (meaning less precise according to Bayesian inference)

What is the problem with compounding uncertainty with each estimate?

It leads to inaccurate predictions.

What is the purpose of updating state from sensory data?

To improve the accuracy of state estimation using Real Robot Dynamics

What are some reasons why sensory data alone may not be sufficient for state estimation?

Failing sensors, noisy data, power and storage constraints.

What is the purpose of the M in SLAM?

To build a map of the local environment using the Kalman Filter

What is LiDAR?

A laser sensor used to measure distance to objects.

What is the purpose of the L in SLAM?

To accurately track the state of the robot.

What is the purpose of extrapolating uncertainty?

To make a prediction about where the robot will be in the next iteration.

What is the purpose of the iterative process in SLAM?

To approximate the true position of the robot.

What does the S in SLAM stand for?

Simultaneous - recursive problem where the robots pose, and its environment estimates are simultaneously refined.

What is the green cloud in SLAM?

Real-time pose

What is the purpose of SLAM?

Localize the robot while mapping the environment

What is the Kalman Filter used for in SLAM?

To recursively build maps and provide insights into robot pose

What does real robot dynamics consist of?

sensor reading = true position(cant observe) + sensor noise(normal distribution)

What is uncertainty?

error estimating a parameter