Skip to main content

A High-Sensitive Green Localization System for High-Speed Self-Driving Vehicles

Periodic Reporting for period 1 - GREENLOC (A High-Sensitive Green Localization System for High-Speed Self-Driving Vehicles)

Okres sprawozdawczy: 2017-08-01 do 2019-07-31

This project is about increasing safety and efficiency in Intelligent Transportation Systems through a novel vehicular localization system GreenLoc. Safer transportation and autonomous driving relies on determining the position and relative velocity of surrounding vehicles and road infrastructures, while energy-efficiency of localization is critical for the sustainability of our planet. The state-of-the-art autonomous driving localization technology suffers from the following serious limitations: 1) Limited maximum speeds when other vehicles exist around due to delay in collecting and processing of location information by the mixed locatlization system of video cameras, LIDAR and radar systems, road markers, ultrasonic sensors, Wi-Fi fingerprinting data, V2V communications through protocols such as WAVE, etc. 2) Ineffectiveness in visual localization due to difficult weather conditions (such as heavy rain and fog) or optical illusions (for example, perceiving a straight road although it bends), stemming from LIDAR and video camera based visual localization of surrounding vehicles. Moreover, energy-efficiency is not provided by the current locatlization systems used in state-of-the-art autonomous vehicles.
This project aims to develop GreenLoc, which is a high-sensitive, fast and green localization platform among vehicles in a multihop vehicular ad-hoc network (VANET).

Crash safety involves taking actions in order to prevent any possible accidents. Accurate localization and determining the position of surrounding vehicles and road-side units with high sensitivity is a necessity for providing crash-safe autonomous vehicles. Reducing delay of localization is also necessary in order to act fast enough before significant position changes occur in presence of high-speed autonomous vehicles. Furthermore, reducing energy costs introduced by the continuous localization process is required for reducing the frequency to charge a high-speed autonomous vehicle, which is the major factor shrinking the average speed. Hence, crash-safe high-speed autonomous vehicles require accurate, fast and energy-efficient localization. Current autonomous vehicle localization technology is insufficient in meeting these three performance measures at the same time, requiring a different approach.

The main goal of this project is providing high-sensitive fast green localization in ITS, serving the ‘Smart, green integrated transport’ focus area of Horizon 2020. GREENLOC aims to localize surrounding vehicles and road-side units (serving instead of conventional traffic lights or stop signs at intersections; or acting as anchors for increasing localization sensitivity in tunnels, closed parking lots and cities), which constitutes a basis for preventing accidents and opening the way to crash-safe high-speed autonomous vehicles.
A joint radar and communication is developed to achieve high-sensitive (15 cm. distance resolution and 1m/s velocity resolution), fast (a delay of at most 100ms) and efficient (spectrum, hardware and energy efficiency) localization in VANETS, while providing a solution to the mutual radar interference problem that threatens safety.

We have quantied under which conditions ghost targets occur and evaluated a joint radar and communication scheme, which reduces interference by adjusting the radar time over a dedicated V2V band, while reusing the radar hardware for communication. By time multiplexing radar transmissions of FMCW automotive radars, we are able to mitigate radar interference and increase pedestrian detection probability without impacting the pedestrian ranging accuracy. Performance in terms of detection probability, SNR, and ranging accuracy are reported, based on high-delity simulations.

The following publications and dissemination activities are accomplished during the project period:
• 2 international (PIMRC and Radar Conference) + 1 national conference (Swedish Transportation Conference)
• 1 journal and 1 magazine article submitted and awaiting decision.
• 1 seminar at WWVC Workshop at Halmstad and seminars given on the subject in the department at Chalmers.
• Participation in Gothenburg Science festival in April 2018 and in MasterPlan seminar at Chalmers in Oct. 2018.
• 1 Master thesis project completed (2018 Sept.)
• Additionally, 2 bachelor theses are completed.
• FFI follow-up funding is attained in cooperation with project partners: Volva Cars Cooperation, Veoneer, QamCom, SAAB, Halmstad University, Chalmers University of Technology, “Combined Radar-Based Communication and Interference Mitigation for AutomotiveApplications”, FFI: Trafiksäkerhet och automatiserade fordon, budget 13.855.000 SEK, 2 year, Jan. 2019-Dec. 2020.
• 1 international patent application is filed.
A novel method for radar interference mitigation and vehicular localization/networking beyond the state-of-art is developed. Hardware demonstration will be accomplished in the follow-up FFI project to test the developed idea. The results of this project might lead to safer and more efficient (spectrum, hardware and energy efficient) vehicular localization and networking for future autonomous driving. Moreover, the filed patent application has the potential to protect the intellectual property produced in scope of this MSCA-IF project by using European resources, which might benefit the European industry in the competitive race of autonomous driving.