Andrew J. Hill

Georgia Markham, Mehala Balamurali, Andrew J. Hill

Few-shot action recognition (FSAR) aims to learn a model capable of identifying novel actions in videos using only a few examples. In assuming the base dataset seen during meta-training and novel dataset used for evaluation can come from different domains, cross-domain few-shot learning alleviates data collection and annotation costs required by methods with greater supervision and conventional (single-domain) few-shot methods. While this form of learning has been extensively studied for image classification, studies in cross-domain FSAR (CD-FSAR) are limited to proposing a model, rather than first understanding the cross-domain capabilities of existing models. To this end, we systematically evaluate existing state-of-the-art single-domain, transfer-based, and cross-domain FSAR methods on new cross-domain tasks with increasing difficulty, measured based on the domain shift between the base and novel set. Our empirical meta-analysis reveals a correlation between domain difference and downstream few-shot performance, and uncovers several important insights into which model aspects are effective for CD-FSAR and which need further development. Namely, we find that as the domain difference increases, the simple transfer-learning approach outperforms other methods by over 12 percentage points, and under these more challenging cross-domain settings, the specialised cross-domain model achieves the lowest performance. We also witness state-of-the-art single-domain FSAR models which use temporal alignment achieving similar or worse performance than earlier methods which do not, suggesting existing temporal alignment techniques fail to generalise on unseen domains. To the best of our knowledge, we are the first to systematically study the CD-FSAR problem in-depth. We hope the insights and challenges revealed in our study inspires and informs future work in these directions.

Rami N. Khushaba, Andrew J. Hill

Radar-based materials detection received significant attention in recent years for its potential inclusion in consumer and industrial applications like object recognition for grasping and manufacturing quality assurance and control. Several radar publications were developed for material classification under controlled settings with specific materials' properties and shapes. Recent literature has challenged the earlier findings on radars-based materials classification claiming that earlier solutions are not easily scaled to industrial applications due to a variety of real-world issues. Published experiments on the impact of these factors on the robustness of the extracted radar-based traditional features have already demonstrated that the application of deep neural networks can mitigate, to some extent, the impact to produce a viable solution. However, previous studies lacked an investigation of the usefulness of lower frequency radar units, specifically <10GHz, against the higher range units around and above 60GHz. This research considers two radar units with different frequency ranges: Walabot-3D (6.3-8 GHz) cm-wave and IMAGEVK-74 (62-69 GHz) mm-wave imaging units by Vayyar Imaging. A comparison is presented on the applicability of each unit for material classification. This work extends upon previous efforts, by applying deep wavelet scattering transform for the identification of different materials based on the reflected signals. In the wavelet scattering feature extractor, data is propagated through a series of wavelet transforms, nonlinearities, and averaging to produce low-variance representations of the reflected radar signals. This work is unique in comparison of the radar units and algorithms in material classification and includes real-time demonstrations that show strong performance by both units, with increased robustness offered by the cm-wave radar unit.

Rami N Khushaba, Arman Melkumyan, Andrew J Hill

Understanding the structure and mineralogical composition of a region is an essential step in mining, both during exploration (before mining) and in the mining process. During exploration, sparse but high-quality data are gathered to assess the overall orebody. During the mining process, boundary positions and material properties are refined as the mine progresses. This refinement is facilitated through drilling, material logging, and chemical assaying. Material type logging suffers from a high degree of variability due to factors such as the diversity in mineralization and geology, the subjective nature of human measurement even by experts, and human error in manually recording results. While laboratory-based chemical assaying is much more precise, it is time-consuming and costly and does not always capture or correlate boundary positions between all material types. This leads to significant challenges and financial implications for the industry, as the accuracy of production blasthole logging and assaying processes is essential for resource evaluation, planning, and execution of mine plans. To overcome these challenges, this work reports on a pilot study to automate the process of material logging and chemical assaying. A machine learning approach has been trained on features extracted from measurement-while-drilling (MWD) data, logged from autonomous drilling systems (ADS). MWD data facilitate the construction of profiles of physical drilling parameters as a function of hole depth. A hypothesis is formed to link these drilling parameters to the underlying mineral composition. The results of the pilot study discussed in this paper demonstrate the feasibility of this process, with correlation coefficients of up to 0.92 for chemical assays and 93% accuracy for material detection, depending on the material or assay type and their generalization across the different spatial regions.

Philip Gun, Andrew Hill, Robin Vujanic

This paper addresses the problem of planning time-optimal trajectories for multiple cooperative agents along specified paths through a static road network. Vehicle interactions at intersections create non-trivial decisions, with complex flow-on effects for subsequent interactions. A globally optimal, minimum time trajectory is found for all vehicles using Mixed Integer Linear Programming (MILP). Computational performance is improved by minimising binary variables using iteratively applied targeted collision constraints, and efficient goal constraints. Simulation results in an open-pit mining scenario compare the proposed method against a fast heuristic method and a reactive approach based on site practices. The heuristic is found to scale better with problem size while the MILP is able to avoid local minima.

Andrew W. Palmer, Robin Vujanic, Andrew J. Hill et al.

The weekly maintenance schedule specifies when maintenance activities should be performed on the equipment, taking into account the availability of workers and maintenance bays, and other operational constraints. The current approach to generating this schedule is labour intensive and requires coordination between the maintenance schedulers and operations staff to minimise its impact on the operation of the mine. This paper presents methods for automatically generating this schedule from the list of maintenance tasks to be performed, the availability roster of the maintenance staff, and time windows in which each piece of equipment is available for maintenance. Both Mixed-Integer Linear Programming (MILP) and genetic algorithms are evaluated, with the genetic algorithm shown to significantly outperform the MILP. Two fitness functions for the genetic algorithm are also examined, with a linear fitness function outperforming an inverse fitness function by up to 5% for the same calculation time. The genetic algorithm approach is computationally fast, allowing the schedule to be rapidly recalculated in response to unexpected delays and breakdowns.

Andrew W. Palmer, Andrew J. Hill, Steven J. Scheding

This paper introduces two objective functions for computing the expected cost in the Stochastic Collection and Replenishment (SCAR) scenario. In the SCAR scenario, multiple user agents have a limited supply of a resource that they either use or collect, depending on the scenario. To enable persistent autonomy, dedicated replenishment agents travel to the user agents and replenish or collect their supply of the resource, thus allowing them to operate indefinitely in the field. Of the two objective functions, one uses a Monte Carlo method, while the other uses a significantly faster analytical method. Approximations to multiplication, division and inversion of Gaussian distributed variables are used to facilitate propagation of probability distributions in the analytical method when Gaussian distributed parameters are used. The analytical objective function is shown to have greater than 99% comparison accuracy when compared with the Monte Carlo objective function while achieving speed gains of several orders of magnitude.

Andrew W. Palmer, Andrew J. Hill, Steven J. Scheding

Consideration of resources such as fuel, battery charge, and storage space, is a crucial requirement for the successful persistent operation of autonomous systems. The Stochastic Collection and Replenishment (SCAR) scenario is motivated by mining and agricultural scenarios where a dedicated replenishment agent transports a resource between a centralised replenishment point to agents using the resource in the field. The agents in the field typically operate within fixed areas (for example, benches in mining applications, and fields or orchards in agricultural scenarios), and the motion of the replenishment agent may be restricted by a road network. Existing research has typically approached the problem of scheduling the actions of the dedicated replenishment agent from a short-term and deterministic angle. This paper introduces a method of incorporating uncertainty in the schedule optimisation through a novel prediction framework, and a branch and bound optimisation method which uses the prediction framework to minimise the downtime of the agents. The prediction framework makes use of several Gaussian approximations to quickly calculate the risk-weighted cost of a schedule. The anytime nature of the branch and bound method is exploited within an MPC-like framework to outperform existing optimisation methods while providing reasonable calculation times in large scenarios.