Plug-and-Sense: A Unified Framework for Self-Calibration of Heterogeneous Sensors in Robotics

• Applicant: Dr. Ernesto Tapia Rodríguez
• Subject Area: Automation, Mechatronics, Control Systems, Intelligent Technical Systems, Robotics ; Image and Language Processing, Computer Graphics and Visualisation, Human Computer Interaction, Ubiquitous and Wearable Computing
• Term: from 2012 to 2015
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 226703374

The objective of the “Plug-and-Sense” project is to develop a computational framework for the automatic self-calibration and coordination of multiple sensors (of different modality) mounted on mobile robots. The framework will include self-calibration methods for singles and multiple cameras, laser range scanners, ultrasonic sensors, infrared cameras, as well as inertial measurements units. Our framework will allow roboticists to mount new sensors in a robot and use the movement of the robot, as well as the measurements from other sensors, to obtain calibrated readings from the new sensors, after only a short adjustment phase. Adding new sensors would become an easy and transparent process. Thus the name of our project: Plug-and-Sense.Most calibration methods used in research or in industry are tedious and time-consuming tasks, requiring expert human intervention, special calibration equipment, and controlled environments in a laboratory. Our research proposal is an effort to provide sensor calibration on-the-fly. Pervasive self-calibration requires autonomous activation of calibration, registration and matching of objects in the environment, and estimation of intrinsic and extrinsic sensor’s parameters. We propose to develop, evaluate, and adapt algorithms generalizing all those steps for cameras and other generic sensors, with the objective of extending self-calibration-principles already used in computer vision to the ability of estimating the parameters of any sensor. Our methods will allow the robot to explore its environment autonomously, immediately after the sensors have been connected.The instrumented cars Made in Germany and E-Instein, as well as a robotic wheel chair, will be made available for the project by FU Berlin. This project will be conducted in parallel to work being done by PhD candidates and undergraduates developing software for the autonomous vehicles. Undergraduate students will conduct the necessary driving tests, in order to check the accuracy and real-time capabilities of the algorithms obtained from this research project. These experiments will be regularly conducted indoors, in the institute’s building and in university halls, as well as outdoors, on streets around the campus.Ideally, the results of this project will make autonomous calibration a fundamental feature of any service robot or robotic vehicle, so that manufacturers can provide Plug-and-Sense components. Thus, the research proposed here would represent a fundamental improvement over the current state of the art in robotics.

• DFG Programme: Research Grants