New frontiers

Developed by the Robotics Institute and Rapido at University of Technology Sydney, Downer and the Rail Manufacturing Cooperative Research Centre (CRC), Dwell Track is a solution using 3D camera technology combined with perception algorithms to anonymously monitor passenger numbers and behavior in real-time on rail station platforms.

Chronic congestion on platforms can lead to extended dwell times when trains stop to put down and pick up passengers. This in turn affects train path capacity, service delivery and reliability. In response, Dwell Track enables station staff to make effective decisions when guiding passengers in real-time and provides long term insights into platform operations around dwell management.

The task of detecting and tracking human movement is challenging, given people are dynamic and come in a variety of shapes, sizes, and appearances. The difficulty of this task is further increased in crowded environments due to frequent visual blockages and the sheer number of people in proximity.

Dwell Track leverages advances in 3D camera technology to extract relevant spatial and temporal information from the rail platform in real-time. Several algorithms identify the head and shoulders, the body parts most visible in a crowd. Dwell Track uses this detail, combined with an understanding of social norms, to track individual movements by determining train door positions, door status, platform occupancy, passenger counts, and the direction passengers are moving.

Dwell Track has received the top Excellence in Innovation award at the CRC Association’s annual conference dinner in 2019. The system has been installed in Wynyard (Train station at centre of Sydney's CBD) in August 2019 to enable Transport for NSW to evaluate the applicability of embedding the technology into the rail operator’s daily operations.

Vehicle localization enables effective remote operation of mining machinery during the cutting process, especially during outburst modes when presence of personnel around the miner is prohibited. A multi-year collaboration between Pempek Systems PTY LTD (An Australian mining equipment OEM) and UTS:RI has revolved around real-time mining machine localization. The collaboration resulted in designing and testing a system to provide real-time information on lateral / longitudinal travel as well as roll/pitch and yaw of miner without the need for modifications to the current operational procedures employed in underground coal mining production. The localization system was integrated into closed loop control of cutting a profile for roadway development with support for testing and integration into miner systems by Elgor and FAMUR in Poland and testing in active production in Borynia coal mine (mine in the south of Poland in Jastrzębie-Zdrój).

Preliminary evaluation of machine localization system was conducted in Tychy, Poland on a REMAG R-2000 Roadheader Miner, initially in an above ground erected tunnel with a concrete poured wall and extreme artificially induced rock/dust and mist conditions. The process of vehicle localization using the bespoke camera-laser-radar system was validated as capable of providing localization information as assistance for operators in sumping into the coalface. Thereafter, the system was integrated with the cutter motions, and on operator demand the system would execute a cut by controlling the cutter to achieve designated profile within tolerances of 20cm (about twice the length of the long edge of a credit card) and allow insertion of ring supports. Thus, demonstrating practical use in roadway development.

Underground testing was performed over the two weeks in Borynia Poland. During the trails in Borynia, facets of the system, with regards to feasibility of all designed processes and accuracy were evaluated in active underground mining.  Overall, with correctly computed localization the system could cut a profile to the designated tolerance, verified from several miner to coalface cutting attempts. During tests, the cutter to coalface interaction did not result in any significant miner position changes that would warrant abandoning the cutting process. The miner position changes that would compromise cutting to specific tolerance were only due to operator motion of the miner in forward/reverse direction once auto-cut already commenced. 

Details on developed system are available from Pempek Pty Ltd (Forced Potato Pty Ltd) or UTS:RI. 

Research Lead/s :  
Dr. Gavin Paul,  
A/Prof. Teresa Vidal Calleja,  
Dr. Marc Carmichael

High Access Localised Operations

This project aims to improve traditional rock scaling operation’s efficiency and safety by utilising Industry 4.0 technologies of robotics, automation, and Virtual Reality (VR). The project’s focus is on delivering a robotic system, with high-level autonomous remote teleoperation, to replace physically demanding manual labour in dangerous working environments. The immersive and intuitive nature of VR will facilitate necessary human interaction, for visualising and examining the proposed system’s sensor data, as well as providing an interface for control and decisions. The project builds upon an ongoing relationship between Ausdrill and UTS, where a scale prototype was collaboratively designed and built in early 2020, and a VR-based digital twin interface developed.

The project’s manufacturing outcome will be the realisation of the developed prototype, its control schemes, and on-site proof-of-concept testing. This platform will be up-scalable as needed in Ausdrill’s Perth manufacturing facility, to become an addition to Ausdrill’s range of deployable products that are manufactured in-house.

Research Strength : Control, planning and coordination, Sensing, perception and estimation, Platforms 

Research Partners : Ausdrill (Perenti Group) 

Research Team : Tony Le, Pham Vu, Nuwan Muhasinghe, Sheilla Sutjipto

Research Lead/s :  
UTS:RI:  
A/Prof. Teresa Vidal Calleja,  
Dr. Marc Carmichael,  
Prof. Robert Fitch  

UTS:CAM:  
Prof. Jochen Deuse,  
Dr, Matthias Guertler,  
Dr. Mickey Clemon

The Australia Cobotics Centre (IC200100001) was established with a $4.9 million grant from the Australian Government through the Australian Research Council’s Industrial Transformation Research Program. The Centre supports Australian manufacturers to shift toward higher-potential markets, compete globally and build a digitally capable workforce for the future. 

The Centre will unite manufacturing businesses and researchers to build the collaborative robotics capability Australian manufacturing industry needs to boost safety, quality assurance, production efficiency, process innovation and workforce readiness. 

A key remit of the Centre is to train researchers, engineers, technologists and manufacturing leaders with the skills and expertise needed to apply collaborative robotics technology in advanced manufacturing contexts, and to manage the human and design factors, business adoption, management quality, and workforce issues that result from implementation of new technology.

Research Strengths : Cobotics, Perception, Human-Robot Interaction, Manipulation, Navigation 

Research Partners : LA Services, B&R Enclosures, Weld Austrlia, IR4, Cook Medical, InfraBuild 

Research Team : QUT (lead), UTS, Swinburne University

Research Lead/s :  
UTS:RI - A/Prof Teresa Vidal Calleja  
DLR -  Dr. Habil. Rudolph Triebel  

Autonomous robots require rich and reliable representations of the environment to interact with the world and perform the designated tasks, in particular, in unknown scenarios such as Planetary-like environments.

Nowadays, several robust real-time navigation strategies for this type of robots still rely only on discrete geometric, sometimes probabilistic, information of the environment, as high contextual representations are in general more expensive to compute and more prone to error. Moreover, commonly extracted semantic representations are seldom probabilistic.

This project aims to investigate the theory and algorithms to create a representation of the environment in continuous space that considers uncertainty and contains semantic information. The key idea is to develop a representation that exploits geometry and, at the same time, context within a probabilistic framework that produces less overconfident semantic information. Such representations will enable the next generation of navigation frameworks for space exploration.

Research Strength : Perception and Navigation 

Research Partners : DLR - German Aerospace Centre 

Research Teams : Cedric Le Gentil, Riccardo Guibilato, Mallikarjuan Vayugundla