PhD Thesis

Robust navigation for industrial service robots

Work default illustration

Information

  • Started: 05/02/2015
  • Finished: 29/09/2020

Description

As one of the fundamental problems of robotics, the different challenges that constitute navigation have been studied for decades. Robust, reliable and safe navigation is a key factor to enable higher level functionalities for robots that are going to evolve around humans. Throughout the present the-sis, we tackle the problem of navigation for robotic industrial mobile-bases. We identify its compo-nents and analyze their respective challenges. The research work presented here ultimately aims at improving the overall quality of the navigation stack of a commercially available industrial mobile-base.
To introduce and survey the overall problem we first break down the navigation framework into clearly identified smaller problems. We examine the problem of simultaneously mapping the envi-ronment and localizing the robot in it by exploring the state of the art. Doing so we recall and detail the mathematical grounding of the Simultaneous Localization and Mapping (SLAM) problem. We then review the problem of planning the trajectory of a mobile-base toward a desired goal in the generated environment representation. Finally we investigate and clarify the concepts and mathe-matical tools of the Lie theory, which we use extensively to provide rigorous mathematical foundation to our developments, focusing on the subset of the theory that is useful to state estimate in robotics.
As the first identified space for improvements, the problem of place recognition for closing loops in SLAM is addressed. Loop closure concerns the ability of a robot to recognize a previously visited loca-tion and infer geometrical information between its current and past locations. Using only a 2D laser range finder sensor, the task is challenging as the perception of the environment provided by the sensor is sparse and limited. We tackle the problem using a technique borrowed from the field of Natural Language Processing (NLP) which has been successfully applied to image-based place recog-nition, namely the Bag-of-Words. We further improve the method with two proposals inspired from NLP. Firstly the comparison of places is strengthen by taking into account the natural relative order of features in each individual sensor readings. Secondly, topological correspondences between places in a corpus of visited places are established in order to promote together instances that are ‘close’ to one another. We evaluate both our proposals separately and jointly on several data sets, with and without noise, and show an improvement over the state of the art.
We then tackle the problem of motion model calibration for odometry estimation. Given a mobile-base embedding an exteroceptive sensor able to observe ego-motion, we propose a novel formula-tion for estimating the intrinsic parameters of an odometry motion model. Resorting to an adaptation of the pre-integration theory initially developed for the IMU motion sensor, we employ iterative non-linear on-manifold optimization to estimate the wheel radii and wheel separation. The method is fur-ther extended to jointly estimate both the intrinsic parameters of the odometry model together with the extrinsic parameters of the embedded sensor. The method is val- idated in simulation and on a real robot and is shown to converge toward the true values of the parameters. It is then shown to accommodate to variation in model parameters quickly when the vehicle is subject to physical chang-es during operation.
Following the generation of a map in which the robot is localized, we address the problem of estimat-ing trajectories for motion planning. We devise a new method for estimating a sequence of robot poses forming a smooth trajectory. Regardless of the Lie group considered, the trajectory is seen as a collection of states lying on a spline with non-vanishing n-th derivatives at every points. Formulated as a multi-objectives nonlinear optimization problem, it allows for the addition of cost functions such as velocity and acceleration limits, collision avoidance and more. The proposed method is evaluated for two different motion planning tasks, the planning of trajectories for a mobile-base evolving in the SE(2) manifold, and the planning of the motion of a multi-link robotic arm whose end-effector evolves in the SE(3) manifold. Furthermore, each task is evaluated in increasingly complex scenarios. In either cases, it is shown to perform comparably or better than the state of the art while producing more consistent results.
From our Lie theory study, we push further the idea of enablement introducing a new, ready to use, programming library called manif. The library is open source, publicly available and is developed fol-lowing software programming good practices. It is designed so that it is easy to integrate and manip-ulate, and allows for flexible use while facilitating the possibility to extend it beyond the already im-plemented Lie groups. Furthermore the library is shown to be efficient compared to other existing solutions.
At last, we come to the conclusion of the doctoral study. We examine the research work and draw lines for future investigations. We also take a look over the past years and share a personal view and experience of the PhD.

The work is under the scope of the following projects:

  • RobInstruct: Instructing robots using natural communication skills (web)
  • LOGIMATIC: Tight integration of EGNSS and on-board sensors for port vehicle automation (web)
  • EB-SLAM: Event-based simultaneous localization and mapping (web)