Developing grounded representations for robots through the principles of sensorimotor coordination

Robots currently recognise and use objects through algorithms that are hand-coded or specifically trained. Such robots can operate in known, structured environments but cannot learn to recognise or use novel objects as they appear. This thesis demonstrates that a robot can develop meaningful object representations by learning the fundamental relationship between action and change in sensory state; the robot learns sensorimotor coordination. Methods based on Markov Decision Processes are experimentally validated on a mobile robot capable of gripping objects, and it is found that object recognition and manipulation can be learnt as an emergent property of sensorimotor coordination.

Classifying natural aerial scenery for autonomous aircraft emergency landing

In this paper, we present an approach for image-based surface classification using multi-class Support Vector Machine (SVM). Classifying surfaces in aerial images is an important step towards an increased aircraft autonomy in emergency landing situations. We design a one-vs-all SVM classifier and conduct experiments on five data sets. Results demonstrate consistent overall performance figures over 88% and approximately 8% more accurate to those published on multi-class SVM on the KTH TIPS data set. We also show per-class performance values by using normalised confusion matrices. Our approach is designed to be executed online using a minimum set of feature attributes representing a feasible and ready-to-deploy system for onboard execution.

Vision based guidance for robot navigation in agriculture

This paper describes a novel vision based texture tracking method to guide autonomous vehicles in agricultural fields where the crop rows are challenging to detect. Existing methods require sufficient visual difference between the crop and soil for segmentation, or explicit knowledge of the structure of the crop rows. This method works by extracting and tracking the direction and lateral offset of the dominant parallel texture in a simulated overhead view of the scene and hence abstracts away crop-specific details such as colour, spacing and periodicity. The results demonstrate that the method is able to track crop rows across fields with extremely varied appearance during day and night. We demonstrate this method can autonomously guide a robot along the crop rows.

Predicting movie ratings from audience behaviors

We propose a method of representing audience behavior through facial and body motions from a single video stream, and use these features to predict the rating for feature-length movies. This is a very challenging problem as: i) the movie viewing environment is dark and contains views of people at different scales and viewpoints; ii) the duration of feature-length movies is long (80-120 mins) so tracking people uninterrupted for this length of time is still an unsolved problem, and; iii) expressions and motions of audience members are subtle, short and sparse making labeling of activities unreliable. To circumvent these issues, we use an infrared illuminated test-bed to obtain a visually uniform input. We then utilize motion-history features which capture the subtle movements of a person within a pre-defined volume, and then form a group representation of the audience by a histogram of pair-wise correlations over a small-window of time. Using this group representation, we learn our movie rating classifier from crowd-sourced ratings collected by rottentomatoes.com and show our prediction capability on audiences from 30 movies across 250 subjects (> 50 hrs).

A convolutional neural network for automatic analysis of aerial imagery

This paper introduces a new method to automate the detection of marine species in aerial imagery using a Machine Learning approach. Our proposed system has at its core, a convolutional neural network. We compare this trainable classifier to a handcrafted classifier based on color features, entropy and shape analysis. Experiments demonstrate that the convolutional neural network outperforms the handcrafted solution. We also introduce a negative training example-selection method for situations where the original training set consists of a collection of labeled images in which the objects of interest (positive examples) have been marked by a bounding box. We show that picking random rectangles from the background is not necessarily the best way to generate useful negative examples with respect to learning.

Score-level multibiometric fusion based on Dempster-Shafer theory incorporating uncertainty factors

While existing multi-biometic Dempster-Shafer the- ory fusion approaches have demonstrated promising perfor- mance, they do not model the uncertainty appropriately, sug- gesting that further improvement can be achieved. This research seeks to develop a unified framework for multimodal biometric fusion to take advantage of the uncertainty concept of Dempster- Shafer theory, improving the performance of multi-biometric authentication systems. Modeling uncertainty as a function of uncertainty factors affecting the recognition performance of the biometric systems helps to address the uncertainty of the data and the confidence of the fusion outcome. A weighted combination of quality measures and classifiers performance (Equal Error Rate) are proposed to encode the uncertainty concept to improve the fusion. We also found that quality measures contribute unequally to the recognition performance, thus selecting only significant factors and fusing them with a Dempster-Shafer approach to generate an overall quality score play an important role in the success of uncertainty modeling. The proposed approach achieved a competitive performance (approximate 1% EER) in comparison with other Dempster-Shafer based approaches and other conventional fusion approaches.

Automated power semiconductor switching performance feature extraction from experimental double-pulse waveform data

Double-pulse tests are commonly used as a method for assessing the switching performance of power semiconductor switches in a clamped inductive switching application. Data generated from these tests are typically in the form of sampled waveform data captured using an oscilloscope. In cases where it is of interest to explore a multi-dimensional parameter space and corresponding result space it is necessary to reduce the data into key performance metrics via feature extraction. This paper presents techniques for the extraction of switching performance metrics from sampled double-pulse waveform data. The reported techniques are applied to experimental data from characterisation of a cascode gate drive circuit applied to power MOSFETs.

What patient information allows us to make accurate predictions of outcome?

Only some of the information contained in a medical record will be useful to the prediction of patient outcome. We describe a novel method for selecting those outcome predictors which allow us to reliably discriminate between adverse and benign end results. Using the area under the receiver operating characteristic as a nonparametric measure of discrimination, we show how to calculate the maximum discrimination attainable with a given set of discrete valued features. This upper limit forms the basis of our feature selection algorithm. We use the algorithm to select features (from maternity records) relevant to the prediction of failure to progress in labour. The results of this analysis motivate investigation of those predictors of failure to progress relevant to parous and nulliparous sub-populations.

Detecting rare events using Kullback-Leibler divergence

One main challenge in developing a system for visual surveillance event detection is the annotation of target events in the training data. By making use of the assumption that events with security interest are often rare compared to regular behaviours, this paper presents a novel approach by using Kullback-Leibler (KL) divergence for rare event detection in a weakly supervised learning setting, where only clip-level annotation is available. It will be shown that this approach outperforms state-of-the-art methods on a popular real-world dataset, while preserving real time performance.

Representing team behaviours from noisy data using player role

Due to their unobtrusive nature, vision-based approaches to tracking sports players have been preferred over wearable sensors as they do not require the players to be instrumented for each match. Unfortunately however, due to the heavy occlusion between players, variation in resolution and pose, in addition to fluctuating illumination conditions, tracking players continuously is still an unsolved vision problem. For tasks like clustering and retrieval, having noisy data (i.e. missing and false player detections) is problematic as it generates discontinuities in the input data stream. One method of circumventing this issue is to use an occupancy map, where the field is discretised into a series of zones and a count of player detections in each zone is obtained. A series of frames can then be concatenated to represent a set-play or example of team behaviour. A problem with this approach though is that the compressibility is low (i.e. the variability in the feature space is incredibly high). In this paper, we propose the use of a bilinear spatiotemporal basis model using a role representation to clean-up the noisy detections which operates in a low-dimensional space. To evaluate our approach, we used a fully instrumented field-hockey pitch with 8 fixed high-definition (HD) cameras and evaluated our approach on approximately 200,000 frames of data from a state-of-the-art real-time player detector and compare it to manually labeled data.

Searching for semantic person queries using channel representations

It is not uncommon to hear a person of interest described by their height, build, and clothing (i.e. type and colour). These semantic descriptions are commonly used by people to describe others, as they are quick to relate and easy to understand. However such queries are not easily utilised within intelligent surveillance systems as they are difficult to transform into a representation that can be searched for automatically in large camera networks. In this paper we propose a novel approach that transforms such a semantic query into an avatar that is searchable within a video stream, and demonstrate state-of-the-art performance for locating a subject in video based on a description.

Supervised latent dirichlet allocation models for efficient activity representation

Local spatio-temporal features with a Bag-of-visual words model is a popular approach used in human action recognition. Bag-of-features methods suffer from several challenges such as extracting appropriate appearance and motion features from videos, converting extracted features appropriate for classification and designing a suitable classification framework. In this paper we address the problem of efficiently representing the extracted features for classification to improve the overall performance. We introduce two generative supervised topic models, maximum entropy discrimination LDA (MedLDA) and class- specific simplex LDA (css-LDA), to encode the raw features suitable for discriminative SVM based classification. Unsupervised LDA models disconnect topic discovery from the classification task, hence yield poor results compared to the baseline Bag-of-words framework. On the other hand supervised LDA techniques learn the topic structure by considering the class labels and improve the recognition accuracy significantly. MedLDA maximizes likelihood and within class margins using max-margin techniques and yields a sparse highly discriminative topic structure; while in css-LDA separate class specific topics are learned instead of common set of topics across the entire dataset. In our representation first topics are learned and then each video is represented as a topic proportion vector, i.e. it can be comparable to a histogram of topics. Finally SVM classification is done on the learned topic proportion vector. We demonstrate the efficiency of the above two representation techniques through the experiments carried out in two popular datasets. Experimental results demonstrate significantly improved performance compared to the baseline Bag-of-features framework which uses kmeans to construct histogram of words from the feature vectors.

Issues in obstacle detection for autonomous mining and construction vehicles

This paper discusses a Dumber of key issues for the development of robust, obstacle detection systems for autonomous mining and construction vehicles. A taxonomy of obstacle detection systems is described; An overview of the state-of- the-art in obstacle detection for outdoor autonomous vehicles is presented with their applicability to the mining and construction environments noted. The issue of so-called fail-safe obstacle detection is then discussed. Finally, we describe the development of an obstacle detection system for a mining vehicle.

Vision-based control for mining automation

The mining industry is highly suitable for the application of robotics and automation technology since the work is both arduous and dangerous. Visual servoing is a means of integrating noncontact visual sensing with machine control to augment or replace operator based control. This article describes two of our current mining automation projects in order to demonstrate some, perhaps unusual, applications of visual servoing, and also to illustrate some very real problems with robust computer vision

Special issue on robot vision

The International Journal of Robotics Research (IJRR) has a long history of publishing the state-of-the-art in the field of robotic vision. This is the fourth special issue devoted to the topic. Previous special issues were published in 2012 (Volume 31, No. 4), 2010 (Volume 29, Nos 2–3) and 2007 (Volume 26, No. 7, jointly with the International Journal of Computer Vision). In a closely related field was the special issue on Visual Servoing published in IJRR, 2003 (Volume 22, Nos 10–11). These issues nicely summarize the highlights and progress of the past 12 years of research devoted to the use of visual perception for robotics.