A5037572995- Robotic Path Planning Algorithms
- Robotics and Sensor-Based Localization
- Optimization and Search Problems
- Reinforcement Learning in Robotics
- Distributed Control Multi-Agent Systems
- Modular Robots and Swarm Intelligence
- 3D Surveying and Cultural Heritage
- Artificial Intelligence in Games
- Topological and Geometric Data Analysis
- Human Pose and Action Recognition
- Robotics and Automated Systems
- Underwater Vehicles and Communication Systems
- Advanced Image and Video Retrieval Techniques
- Advanced Data Processing Techniques
- Neural Networks and Applications
- Fault Detection and Control Systems
- Remote Sensing and LiDAR Applications
- Robotic Locomotion and Control
- Transportation and Mobility Innovations
- Vehicle Routing Optimization Methods
- Smart Parking Systems Research
- Metaheuristic Optimization Algorithms Research
- Smart Agriculture and AI
- Teleoperation and Haptic Systems
- Advanced Vision and Imaging
University of Technology Sydney
2019-2025
Oregon State University
2018-2023
Australian Centre for Robotic Vision
2015-2020
The University of Sydney
2015-2020
Engineering Systems (United States)
2013
United Kingdom Atomic Energy Authority
1967-1968
We propose a decentralized variant of Monte Carlo tree search (MCTS) that is suitable for variety tasks in multi-robot active perception. Our algorithm allows each robot to optimize its own actions by maintaining probability distribution over plans the joint-action space. Robots periodically communicate compressed form their trees, which are used update joint using distributed optimization approach inspired variational methods. method admits any objective function defined action sequences,...
Subterranean robot exploration is difficult, with many mobility, communications, and navigation challenges that require an approach a diverse set of systems, reliable autonomy. While prior work has demonstrated partial successes in addressing the problem, here we convey comprehensive to address problem subterranean wide range tunnel, urban, cave environments. Our driven by themes resiliency modularity, show examples how these influence design different modules. In particular, detail our...
Contextual cues can provide a rich source of information for robots that operate in the presence other agents such as people, animals, vehicles and fellow robots. We are interested context, form behavioural intent an agent, enhanced trajectory prediction. present Bayesian framework estimates both intended goal destination future mobile agent moving among multiple static obstacles. Our method is based on multi-modal hypotheses goal, focused primarily long-term agent. propose computationally...
We present an algorithm for selecting when to communicate during online planning phases of coordinated multi-robot missions. The key idea is that a robot decides request communication from another by reasoning over the predicted information value messages sliding time-horizon, where are probability distributions action sequences. formulate this problem in context recently proposed decentralised Monte Carlo tree search (Dec-MCTS) online, coordination. propose particle filter predicting value,...
We consider the problem of reconstructing regions interest a scene using multiple robot arms and RGB-D sensors.This is motivated by variety applications, such as precision agriculture infrastructure inspection.A viewpoint evaluation function presented that exploits predicted observations geometry scene.A recently proposed non-myopic planning algorithm, Decentralised Monte Carlo tree search, used to coordinate actions arms.Motion performed over navigation graph considers high-dimensional...
Abstract Ocean monitoring is an expensive and time consuming endeavor, but it can be made more efficient through the use of teams autonomous robots. In this paper, we present a system for identification tracking ocean fronts by coordinating sampling efforts heterogeneous team surface vehicles (ASVs) underwater (AUVs). The primary contributions study are (1) our algorithm performing coordination using general autonomy principles: Sequential Allocation Monte Carlo Tree Search (SA‐MCTS) which...
We propose a self-organising map (SOM) algorithm as solution to new multi-goal path planning problem for active perception and data collection tasks. optimise paths multi-robot team that aims maximally observe set of nodes in the environment. The selected are observed by visiting associated viewpoint regions defined sensor model. key characteristics overlapping polygonal continuous regions, each node has an observation reward, robots constrained travel budgets. SOM jointly selects allocates...
In this paper, we address the Orienteering problem (OP) by unsupervised learning of self-organizing map (SOM). We propose to solve OP with a new algorithm based on SOM for Traveling salesman (TSP). Both problems are similar in finding tour visiting given locations; however, stands determine most valuable that maximizes rewards collected subset locations while keeping length under specified travel budget. The proposed stochastic search is and it constructs feasible solution during each epoch....
We present a coordinated autonomy pipeline for multi-sensor exploration of confined environments.We simultaneously address four broad challenges that are typically overlooked in prior work: (a) make effective use both range and vision sensing modalities, (b) perform this across wide environments, (c) be resilient to adverse events, (d) execute onboard team physical robots.Our solution centers around behavior tree architecture, which adaptively switches between various behaviors involving...
We consider an optimal stopping formulation of the mission monitoring problem, in which a monitor vehicle must remain close proximity to autonomous robot that stochastically follows predicted trajectory. This problem arises diverse range scenarios, such as underwater vehicles supervised by surface vessels, pedestrians monitored aerial vehicles, and animals agricultural robots. The key characteristics we are stationary while observing robot, motion is modeled general stochastic process,...
Robotic exploration requires adaptively selecting navigation goals that result in the rapid discovery and mapping of an unknown world. In many real-world environments, subtle structural cues can provide insight about unexplored world, which may be exploited by a decision maker to improve speed exploration. sparse subterranean tunnel networks, these come form topological features, such as loops or dead-ends, are often common across similar environments. We propose method for learning features...
We address the problem of generating navigation roadmaps for uncertain and cluttered environments represented with probabilistic occupancy maps. A key challenge is to generate that provide connectivity through tight passages paths around obstacles. propose topology-informed growing neural gas algorithm leverages estimates topological structures computed using persistent homology theory. These structure inform random sampling distribution focus roadmap learning on challenging regions...
We present an algorithm for learning behavior trees robotic task planning, which alleviates the need time-intensive or infeasible manual design of control architectures. Our method involves representing search space as a formal grammar and searching over this by means new generalization Monte Carlo tree (MCTS) directed acyclic graphs (DAGs), named MCDAGS. Additionally, our employs simulated annealing to expedite aggregation most functional subtrees. experiments marine target response...
We present a coordinated autonomy pipeline for multi-sensor exploration of confined environments. simultaneously address four broad challenges that are typically overlooked in prior work: (a) make effective use both range and vision sensing modalities, (b) perform this across wide environments, (c) be resilient to adverse events, (d) execute onboard teams physical robots. Our solution centers around behavior tree architecture, which adaptively switches between various behaviors involving...
This paper presents a new coordination algorithm for decentralised multi-robot information gathering. We consider planning an online variant of the multi-agent orienteering problem with neighbourhoods. formulation closely aligns number important tasks in robotics, including inspection, surveillance, and reconnaissance. propose self-organising map (SOM) learning procedure, named Dec-SOM, which efficiently plans sequences waypoints team robots. Decentralisation is achieved by performing...
We consider an optimal stopping formulation of the mission monitoring problem, where a monitor vehicle must remain in close proximity to autonomous robot that stochastically follows pre-planned trajectory.This problem arises when underwater vehicles are monitored by surface vessels, and diverse range other scenarios.The key characteristics we stationary while observing robot, motion is modelled general as stochastic process.We propose resolution-complete algorithm for this runs polynomial...
Autonomous robot navigation in austere environments is critical to missions like “search and rescue”, yet it remains difficult achieve. The recent DARPA Subterranean Challenge (SubT) highlights prominent challenges including GPS-denied through rough terrains, rapid exploration large-scale three-dimensional (3D) space, interrobot coordination over unreliable communication. Solving these requires both mechanical resilience algorithmic intelligence. Here, we present our approach that leverages...
This extended abstract presents a new coordination algorithm for decentralised multi-robot information gathering. We consider planning an online variant of the multi-agent orienteering problem with neighbourhoods. formulation closely aligns number important tasks in robotics, including inspection, surveillance, and reconnaissance. propose self-organising map (SOM) learning procedure, named Dec-SOM, which efficiently finds plans team robots non-myopic manner. Decentralisation is achieved by...
We present a new algorithm for deploying passenger robots in marsupial robot systems. A system consists of carrier (e.g., ground vehicle), which is highly capable and has long mission duration, at least one short-duration aerial vehicle) transported by the carrier. optimize performance deployment proposing an that reasons over uncertainty exploiting information about prior probability distribution features interest environment. Our formulated as solution to sequential stochastic assignment...