Virgona, A, Alempijevic, A & Vidal-Calleja, T 2018, 'Socially constrained tracking in crowded environments using shoulder pose estimates', Proceedings - IEEE International Conference on Robotics and Automation, IEEE International Conference on Robotics and Automation, IEEE, Brisbane, QLD, Australia, pp. 4555-4562.View/Download from: UTS OPUS or Publisher's site
© 2018 IEEE. Detecting and tracking people is a key requirement in the development of robotic technologies intended to operate in human environments. In crowded environments such as train stations this task is particularly challenging due the high numbers of targets and frequent occlusions. In this paper we present a framework for detecting and tracking humans in such crowded environments in terms of 2D pose (x, y, θ). The main contributions are a method for extracting pose from the most visible parts of the body in a crowd, the head and shoulders, and a tracker which leverages social constraints regarding peoples orientation, movement and proximity to one another, to improve robustness in this challenging environment. The framework is evaluated on two datasets: one captured in a lab environment with ground truth obtained using a motion capture system, and the other captured in a busy inner city train station. Pose errors are reported against the ground truth and the tracking results are then compared with a state-of-the-art person tracking framework.
Virgona, A, Kirchner, N & Alempijevic, A 2015, 'Sensing and Perception Technology to Enable Real Time Monitoring of Passenger Movement Behaviours Through Congested Rail Stations', 2015 ATRF Conference Proceeding, Australasian Transport Research Forum, ATRF, Sydney.View/Download from: UTS OPUS
The real time monitoring of passenger movement and behaviour through public transport environments including precincts, concourses, platforms and train vestibules would enable operators to more effectively manage congestion at a whole-of-station level. While existing crowd monitoring technologies allow operators to monitor crowd densities at critical locations and react to overcrowding incidents, they do not necessarily provide an understanding of the cause of such issues. Congestion is a complex phenomenon involving the movements of many people though a set of spaces and monitoring these spaces requires tracking large numbers of individuals. To do this, traditional surveillance technologies might be used but at the expense of introducing privacy concerns. Scalability is also a problem, as complete sensor coverage of entire rail station precinct, concourse and platform areas potentially requires a high number of sensors, increasing costs. In light of this, there is a need for sensing technology that collects data from a set of 'sparse sensors', each with a limited field of view, but which is capable of forming a network that can track the movement and behaviour of high numbers of associated individuals in a privacy sensitive manner. This paper presents work towards the core crowd sensing and perception technology needed to enable such a capability. Building on previous research using three-dimensional (3D) depth camera data for person detection, a privacy friendly approach to tracking and recognising individuals is discussed. The use of a head-to-shoulder signature is proposed to enable association between sensors. Our efforts to improve the reliability of this measure for this task are outlined and validated using data captured at Brisbane Central rail station.
Kirchner, N, Alempijevic, A, Virgona, A, Dai, X, Ploger, PG & Venkat, RK 2014, 'A robust people detection, tracking, and counting system', Proceedings of the Australasian Conference on Robotics and Automation - A robust people detection, tracking, and counting system, Australasian Conference on Robotics and Automation, Australasian Robotics and Automation Association, Melbourne, Australia, pp. 1-8.View/Download from: UTS OPUS
The ability to track moving people is a key aspect of autonomous robot system in real-world environments. Whilst for many tasks knowing the approximate positions of people may be sufficient, the ability to identify unique people is needed to accurately count the people in real world. To accomplish the people counting task, a robust system in people detection, tracking and identification is needed.
This paper presents our approach for robust real world people detection, tracking and counting using a PrimeSense RGBD camera. Our past research, upon which we built, is highlighted and novel methods to solve the problems of sensors self localization, false negatives due to persons physically interacting with the environment, and track misassociation due to crowdedness are presented.
An empirical evaluation of our approach in a major Sydney public train station N=420 was conducted, and results demonstrating our methods in the complexities of this challenging environment are presented.
Kirchner, NG, Alempijevic, A & Virgona, A 2012, 'Head-To-Shoulder Signature for Person Recognition', Robotics and Automation (ICRA), 2012 IEEE International Conference on, IEEE International Conference on Robotics and Automation, IEEE, St Paul, MN, USA, pp. 1226-1231.View/Download from: UTS OPUS or Publisher's site
Ensuring that an interaction is initiated with a particular and unsuspecting member of a group is a complex task. As a first step the robot must effectively, expediently and reliably recognise the humans as they carry on with their typical behaviours (in situ). A method for constructing a scale and viewing angle robust feature vector (from analysing a 3D pointcloud) designed to encapsulate the inter-person variations in the size and shape of the people's head to shoulder region (Head-to-shoulder signature - HSS) is presented. Furthermore, a method for utilising said feature vector as the basis of person recognition via a Support-Vector Machine is detailed. An empirical study was performed in which person recognition was attempted on in situ data collected from 25 participants over 5 days in a office environment. The results report a mean accuracy over the 5 days of 78.15% and a peak accuracy 100% for 9 participants. Further, the results show a considerably better-than-random (1/23 = 4.5%) result for when the participants were: in motion and unaware they were being scanned (52.11%), in motion and face directly away from the sensor (36.04%), and post variations in their general appearance. Finally, the results show the HSS has considerable ability to accommodate for a person's head, shoulder and body rotation relative to the sensor - even in cases where the person is faced directly away from the robot.