Robust Estimation and Applications in Robotics

Solving estimation problems is a fundamental component of numerousrobotics applications. Prominent examples involve pose estimation, pointcloud alignment, or object tracking. Algorithms for solving these estimationproblems need to cope with new challenges due to an increased use of potentiallypoor low-cost sensors, and an ever growing deployment of roboticalgorithms in consumer products which operate in potentially unknown environments.These algorithms need to be capable of being robust against strongnonlinearities, high uncertainty levels, and numerous outliers. However, particularlyin robotics, the Gaussian assumption is prevalent in solutions to multivariateparameter estimation problems without providing the desired level ofrobustness.The goal of this tutorial is helping to address the aforementioned challengesby providing an introduction to robust estimation with a particularfocus on robotics. First, this is achieved by giving a concise overview of thetheory on M-estimation. M-estimators share many of the convenient propertiesof least-squares estimators, and at the same time are much more robustto deviations from the Gaussian model assumption. Second, we present severalexample applications where M-Estimation is used to increase robustnessagainst nonlinearities and outliers.