Adding a receding horizon to Locally Weighted Regression for learning robot control

There have been notable advances in learning to control complex robotic systems using methods such as Locally Weighted Regression (LWR). In this paper we explore some potential limits of LWR for robotic applications, particularly investigating its application to systems with a long horizon of temporal dependence. We define the horizon of temporal dependence as the delay from a control input to a desired change in output. LWR alone cannot be used in a temporally dependent system to find meaningful control values from only the current state variables and output, as the relationship between the input and the current state is under-constrained. By introducing a receding horizon of the future output states of the system, we show that sufficient constraint is applied to learn good solutions through LWR. The new method, Receding Horizon Locally Weighted Regression (RH-LWR), is demonstrated through one-shot learning on a real Series Elastic Actuator controlling a pendulum.

Learning multidimensional joint control of a robot using receding horizon locally weighted regression

In this paper we explore the ability of a recent model-based learning technique Receding Horizon Locally Weighted Regression (RH-LWR) useful for learning temporally dependent systems. In particular this paper investigates the application of RH-LWR to learn control of Multiple-input Multiple-output robot systems. RH-LWR is demonstrated through learning joint velocity and position control of a three Degree of Freedom (DoF) rigid body robot.

The RatSLAM project : robot spatial navigation

Rats are superior to the most advanced robots when it comes to creating and exploiting spatial representations. A wild rat can have a foraging range of hundreds of meters, possibly kilometers, and yet the rodent can unerringly return to its home after each foraging mission, and return to profitable foraging locations at a later date (Davis, et al., 1948). The rat runs through undergrowth and pipes with few distal landmarks, along paths where the visual, textural, and olfactory appearance constantly change (Hardy and Taylor, 1980; Recht, 1988). Despite these challenges the rat builds, maintains, and exploits internal representations of large areas of the real world throughout its two to three year lifetime. While algorithms exist that allow robots to build maps, the questions of how to maintain those maps and how to handle change in appearance over time remain open. The robotic approach to map building has been dominated by algorithms that optimise the geometry of the map based on measurements of distances to features. In a robotic approach, measurements of distance to features are taken with range-measuring devices such as laser range finders or ultrasound sensors, and in some cases estimates of depth from visual information. The features are incorporated into the map based on previous readings of other features in view and estimates of self-motion. The algorithms explicitly model the uncertainty in measurements of range and the measurement of self-motion, and use probability theory to find optimal solutions for the geometric configuration of the map features (Dissanayake, et al., 2001; Thrun and Leonard, 2008). Some of the results from the application of these algorithms have been impressive, ranging from three-dimensional maps of large urban strucutures (Thrun and Montemerlo, 2006) to natural environments (Montemerlo, et al., 2003).

Gaussian process models for sensor-centric robot localisation

This paper presents an approach to building an observation likelihood function from a set of sparse, noisy training observations taken from known locations by a sensor with no obvious geometric model. The basic approach is to fit an interpolant to the training data, representing the expected observation, and to assume additive sensor noise. This paper takes a Bayesian view of the problem, maintaining a posterior over interpolants rather than simply the maximum-likelihood interpolant, giving a measure of uncertainty in the map at any point. This is done using a Gaussian process framework. To validate the approach experimentally, a model of an environment is built using observations from an omni-directional camera. After a model has been built from the training data, a particle filter is used to localise while traversing this environment

Marking out the pitch : Defining the soccer fiction genre

The fiction of football, or soccer as it is commonly known in both the US and Australia, has a long and deep history. Despite this, there is still academic uncertainty as to whether it qualifies description as a distinct fictive genre. This paper seeks to address this question. It brings a genre analysis approach to the football fiction canon. Starting with definitions of what constitutes football fiction and genre, the paper goes on to distinguish and aggregate common trends, tensions and divergences and identify emerging and popular movements within football fiction. The paper will then assess football fiction’s status as a discrete genre.

Interpersonal coordination tendencies shape 1-vs-1 sub-phase performance outcomes in youth soccer

This study investigated the influence of interpersonal coordination tendencies on performance outcomes of 1-vs-1 subphases in youth soccer. Eight male developing soccer players (age: 11.8+0.4 years; training experience: 3.6+1.1 years) performed an in situ simulation of a 1-vs-1 sub-phase of soccer. Data from 82 trials were obtained with motion-analysis techniques, and relative phase used to measure the space-time coordination tendencies of attacker-defender dyads. Approximate entropy (ApEn) was then used to quantify the unpredictability of interpersonal interactions over trials. Results revealed how different modes of interpersonal coordination emerging from attacker-defender dyads influenced the 1-vs-1 performance outcomes. High levels of space-time synchronisation (47%) and unpredictability in interpersonal coordination processes (ApEn: 0.91+0.34) were identified as key features of an attacking player’s success. A lead-lag relation attributed to a defending player (34% around 7308 values) and a more predictable coordination mode (ApEn: 0.65+0.27, P50.001), demonstrated the coordination tendencies underlying the success of defending players in 1-vs-1 sub-phases. These findings revealed how the mutual influence of each player on the behaviour of dyadic systems shaped emergent performance outcomes. More specifically, the findings showed that attacking players should be constrained to exploit the space-time synchrony with defenders in an unpredictable and creative way, while defenders should be encouraged to adopt postures and behaviours that actively constrain the attacker’s actions.

Sophoclean soccer

This piece emerged during the development of my PhD research. I started looking at the differences between young adult and adult fiction. I found strangebOUnce.com to be kindred spirits looking at football, the beautiful game, through creators eyes. I hope to publish more work with them

Dynamic robot path planning using an enhanced simulated annealing approach

Evolutionary computation is an effective tool for solving optimization problems. However, its significant computational demand has limited its real-time and on-line applications, especially in embedded systems with limited computing resources, e.g., mobile robots. Heuristic methods such as the genetic algorithm (GA) based approaches have been investigated for robot path planning in dynamic environments. However, research on the simulated annealing (SA) algorithm, another popular evolutionary computation algorithm, for dynamic path planning is still limited mainly due to its high computational demand. An enhanced SA approach, which integrates two additional mathematical operators and initial path selection heuristics into the standard SA, is developed in this work for robot path planning in dynamic environments with both static and dynamic obstacles. It improves the computing performance of the standard SA significantly while giving an optimal or near-optimal robot path solution, making its real-time and on-line applications possible. Using the classic and deterministic Dijkstra algorithm as a benchmark, comprehensive case studies are carried out to demonstrate the performance of the enhanced SA and other SA algorithms in various dynamic path planning scenarios.

One robot, eight hours, and twenty four thousand people

This paper presents an account of an autonomous mobile robot deployment in a densely crowded public event with thousands of people from different age groups attending. The robot operated for eight hours on an open floor surrounded by tables, chairs and massive touchscreen displays. Due to the large number of people who were in close vicinity of the robot, different safety measures were implemented including the use of no-go zones which prevent the robot from blocking emergency exits or moving too close to the display screens. The paper presents the lessons learnt and experiences obtained from this experiment, and provides a discussion about the state of mobile service robots in such crowded environments.

Prediction of muscular architecture of the rectus femoris and vastus lateralis from EMG during isometric contractions in soccer players

Objectives The purpose of the study was to establish regression equations that could be used to predict muscle thickness and pennation angle at different intensities from electromyography (EMG) based measures of muscle activation during isometric contractions. Design Cross-sectional study. Methods Simultaneous ultrasonography and EMG were used to measure pennation angle, muscle thickness and muscle activity of the rectus femoris and vastus lateralis muscles, respectively, during graded isometric knee extension contractions performed on a Cybex dynamometer. Data form fifteen male soccer players were collected in increments of approximately 25% intensity of the maximum voluntary contraction (MVC) ranging from rest to MVC. Results There was a significant correlation (P < 0.05) between ultrasound predictors and EMG measures for the muscle thickness of rectus femoris with an R2 value of 0.68. There was no significant correlation (P > 0.05) between ultrasound pennation angle for the vastus lateralis predictors for EMG muscle activity with an R2 value of 0.40. Conclusions The regression equations can be used to characterise muscle thickness more accurately and to determine how it changes with contraction intensity, this provides improved estimates of muscle force when using musculoskeletal models.

Online belongings : female fan experiences in online soccer forums

Introduction: The Paradox at the Heart of Online Interaction

Autonomous movement-driven place recognition calibration for generic multi-sensor robot platforms

In this paper we present a method for autonomously tuning the threshold between learning and recognizing a place in the world, based on both how the rodent brain is thought to process and calibrate multisensory data and the pivoting movement behaviour that rodents perform in doing so. The approach makes no assumptions about the number and type of sensors, the robot platform, or the environment, relying only on the ability of a robot to perform two revolutions on the spot. In addition, it self-assesses the quality of the tuning process in order to identify situations in which tuning may have failed. We demonstrate the autonomous movement-driven threshold tuning on a Pioneer 3DX robot in eight locations spread over an office environment and a building car park, and then evaluate the mapping capability of the system on journeys through these environments. The system is able to pick a place recognition threshold that enables successful environment mapping in six of the eight locations while also autonomously flagging the tuning failure in the remaining two locations. We discuss how the method, in combination with parallel work on autonomous weighting of individual sensors, moves the parameter dependent RatSLAM system significantly closer to sensor, platform and environment agnostic operation.