Active Vision Methods in Autonomous Navigation

Tuesday, June 28, 2022, 9:00 AM – 5:00 PM

Organizers

Dr.  Konstantinos Tsintotas, Democritus University of Thrace, Greece

Dr. Nitin J. Sanket, University of Maryland, College Park, USA

Prof. Antonios Gasteratos, Democritus University of Thrace, Greece

Prof. Yiannis Aloimonos, University of Maryland, College Park, USA

Workshop Summary

In applications involving deployment of mobile robots, it is a fundamental requirement for the robot to perceive its environment and to decide about its ability to navigate, as well as how to navigate in general.  Currently, every mobile robot is equipped with on-board sensors that contribute to observing the working environment and to collect and process data. However, when any mobile robot starves from computational and sensing resources, simple approaches using passive perception, such as building a 3D map of the environment, fall apart. It is noted that active vision, which involves building algorithmic engines around the perception-action synergy loop, offers a promising solution as it allows for the mobile robot to actively use its sensors towards specific viewpoints according to a specific scanning strategy. Thus, the challenge in active vision systems is that the robot must decide “where to look” following a plan. Thus, vision sensors must be purposefully configured and placed at several positions with the aim to observe a target by taking actions linked with robotic perception, which, in turn, introduces active and purposeful behaviors. In particular, four modes of activeness have been formally identified: By moving the agent itself; By employing an active sensor; By moving a part of the agent’s body; By hallucinating active movements. The Workshop covers several challenging areas including Autonomous navigation; Perception-Action coupling; Driverless cars; UAVs; SWAP-Aware design; Semantic mapping and navigation; SLAM; Loop-closure detection; Visual odometry.

Schedule

09:00 – 09:30
Welcome
09:30 – 10:30Invited talk: Learning to Fly: From perception to action,
Prof. Davide Scaramuzza, University of Zurich, Switzerland
10:30 – 11:00Coffee Break
11:00 – 12:00Invited talk: Resilient Robotic Autonomy – Technologies and Lessons Learned from the DARPA Subterranean Challenge,
Prof. Kostas Alexis, Norwegian University of Science and Technology, Norway.
12:00 – 13:00Paper presentations
13:00 – 14:00Lunch break
14:00 – 15:00
Invited talk: Active Perception with Foveal Sensors: a bio-inspired strategy for resource constrained robots,
Prof. Alexandre Bernardino, Instituto Superior Técnico, Portugal
15:00 -16:00
Invited talk: Navigation During Visuospatial Problem Solving,
Prof. John K. Tsotsos, York University, Canada
16:00 – 16:30Coffee Break
16:30 – 17:00Panel Discussion and Concluding Remarks