Ad Usum Navigantium (AD USUM)

TETRA project on sensor fusion and autonomous map building

Interested in collaboration on this topic? Please contact us via e-mail ( eric.demeester@kuleuven.be ) or phone ( +32(0)11 75.17.75 ).

More information at: https://iiw.kuleuven.be/onderzoek/acro/adusumnavigantium

Context

With the rise of mobile robots in our daily lives such as robotic vacuum cleaners, robotic lawn mowers, drones and self-driving cars, it is clear that navigation in unstructured environments is becoming increasingly important. This is also noticeable with the emergence of new applications for autonomous navigation in shipping, agriculture, the nuclear sector, food, pharmaceutics industry, etc. The first necessary step to achieve collision-free navigation in such environments is to estimate the robot location and the geometry of the environment. The robot location can be determined based on different sensor technologies such as GNSS or laser, but unfortunately no sensor works flawlessly under all circumstances. A lot of subcentimetric localization technology also uses a map of the environment; that map must also be determined. It is therefore important to optimally combine the right sensor technologies in order to construct a map of the environment in a robust way and at the same time determine the robot location in this map; in robotics literature this is called SLAM: Simultaneous Localization and Mapping. This TETRA project aims to transfer state-of-the-art SLAM technology to Flemish industry in order to improve their competitive position.

Key results

1. A literature study and market study on indoor positioning systems and map building software;
2. SLAM related validations in laboratory conditions, including an evaluation of the Marvelmind ultrasonic localization system, a benchmarking of existing 2D SLAM software, an experimental comparison of 4 different, commercially available IMUs (Inertial Measurement Units), an experimental comparison of different indoor positioning systems and an algorithm and software for time synchronization of cheap cameras and IMUs, using visual odometry.
3. Interfacing and evaluation of relevant sensors (2D laser scanners, 3D cameras and laser scanners, IMUs, GNSS); the majority of these sensors is still available in our lab for experimental tests and projects in collaboration with industrial partners.
4. Development of 2 mobile sensor platforms for use in case studies and for didactic purposes.
5. Case studies in the field: indoor cases on the mapping of buildings, and outdoor cases on the mapping of an orchard and inland shipping environments.
6. Scientific publications and dissemination activities (including a two-day demonstration at Digital Construction 2018).
7. Master theses in close collaboration with industry.