AdMaLL: Advanced 3d Mapping and Long-term Localisation with semantics understanding and terrain assessment for robot navigation in gps-denied environments
Principal Investigator
Co-PI
Dr.Jaime Pulido Fentanes
Co-PI (Industrial)
Advisory board
working on multi-sensor perception, mapping
Ahmed Abbas
working on robot localisation
Christopher Wild
working on terrain assessment
Jiahui Liu
working on anomalous object detection
Weihang Zhang
working on visual based navigation
Zhicheng Zhou
working on lidar-based global localisation
News - Sep 2021 - IROS presentation of V-T&R navigation paper
Paper: [PDF]
News-July 2021 - Field testing on West Wittering beach
We have tested our mapping, localization, and navigation system on Mars-like scene (The West Wittering beach).
News - April 2021 - Submap and MCL testing
In this video, the multi-sensor mapper and NDT-MCL localsiation are tested in Sheffield campus. Failure cases are detected and fixed. The system is ready for field test.
Quarterly Deliverable - Q4 [PDF]
News - February 2021 - Annual review video is released
Annual review demo videos are delivered to the FAIR-SPACE Hub.
News - January 2021 - Map-based relocalisation is integrated
We have integrated NDT-map representation and NDT-MCL (Monte-Carlo Localisation) with submap implementation with the Jaguar platform. The robot is able to re-localise (in 6DOF) within a previous-build map robustly and smoothly in large-scale unstructured environments.
Task 2.2, and the major part of 2.3 (NDT-map implementation, NDT-MCL integration) are completed. Note Semantic mapping in Task 2.3 is under development.
Quarterly Deliverable - Q3 [PDF]
News - November 2020 - Visual-based navigation stack is released.
We released the navigation stack for long-term visual navigation. Our approach is able to operate from-day-to-night using a single map and can repeat complex manoeuvres.
Github: https://github.com/kevinlisun/jaguar_nav
News - September 2020 - Multi-sensor mapping demo.
Task 1.1, 1.2, 2.1 (multi-sensor installation, calibration, multi-sensor mapping) are completed.
Quarterly Deliverable - Q2 [PDF]
News - September 2020 - Second field test and data collection
Visual-based navigation with various terrains. The second field test at Longshaw National Trust, Sheffield, UK.
News - August 2020 - First field test
Visual-based navigation in unstructured environments. The first field test at Tankersley, Sheffield, UK.
News - August 2020 - new demo of visual navigation
The low-level control and teleoperation is implemented. The visual navigation stack is integrated with ZED camera and Jaguar robot. Autonomous visual teach repeat (monocular and appearane-only) is demostrated in Romanwharf, Lincoln.
News - June 2020 - CAD model is finalised
The hardware design of the multi-sensor perception system is finalised. Our perception system consists of an Ouster64 lidar, a ZED2 stereo camera, Xsens 750 IMU.
Quarterly Deliverable - Q1 [PDF]
News - Jan 2020 - new publication
Our robot localisation research is presented in ICRA 2020.
The proposed approach is able to localise the robot using a singal lidar sensor within 2 seconds in average in a large-scale environment (0.5 km2). [Presentation]