Implementation of semi-active damping on a tri-axle heavy-vehicle suspension

A semi-active truck damper was developed in conjunction with a commercial shock absorber manufacturer. A linearized damper model was developed for control system design purposes. Open- and closed-loop damper force tracking control was implemented, with tests showing that an open-loop approach gave the best compromise between response speed and accuracy. A hardware-in-the-loop test facility was used to investigate performance of the damper when combined with a simulated quarter-car model. The input to the vehicle model was a set of randomly generated road profiles, each profile traversed at an appropriate speed. Modified skyhook damping tests showed a simultaneous improvement over the optimum passive case of 13 per cent in vertical body acceleration and 8 per cent in dynamic tyre forces. Full-scale vehicle tests of the damper on a heavy tri-axle trailer were carried out. Implementation of modified skyhook damping yielded a simultaneous improvement over the optimum passive case of 8 per cent in vertical body acceleration and 8 per cent in dynamic tyre forces. © IMechE 2008.

An experimental semi-active anti-roll system

The use of anti-roll bars to provide additional roll stiffness and therefore to reduce the trade-off between ride and rollover performance has previously been studied. However, little work has been carried out to investigate the benefits of a switchable roll stiffness. Such a semi-active anti-roll system has the ability to have a low roll stiffness during straight-ahead driving for improved ride performance and high roll stiffness during cornering for improved roll performance. Modelling of such a system is conducted and the model is validated against a semi-active anti-roll system fitted to an experimental vehicle. Experimental and theoretical investigations are used to investigate the performance of such a system with several different strategies employed to switch to the high-stiffness state. The use of an air suspension on the vehicle to roll into corners is also investigated, as is the possibility of exploiting the road layout by allowing the vehicle to be in a low-roll-stiffness configuration during a corner, and then to switch to the high-roll-stiffness configuration midcorner, hence 'locking in' a roll angle. The best rollover performance improvement that was achieved was 12.5 per cent. © IMechE 2008.

Decision tree-based context clustering based on cross validation and hierarchical priors

The standard, ad-hoc stopping criteria used in decision tree-based context clustering are known to be sub-optimal and require parameters to be tuned. This paper proposes a new approach for decision tree-based context clustering based on cross validation and hierarchical priors. Combination of cross validation and hierarchical priors within decision tree-based context clustering offers better model selection and more robust parameter estimation than conventional approaches, with no tuning parameters. Experimental results on HMM-based speech synthesis show that the proposed approach achieved significant improvements in naturalness of synthesized speech over the conventional approaches. © 2011 IEEE.

Active steering of a tractor-semi-trailer

This paper develops a path-following steering control strategy for an articulated heavy goods vehicle. The controller steers the axles of the semi-trailer so that its rear end follows the path of the fifth wheel coupling: for all paths and all speeds. This substantially improves low-speed manoeuvrability, off-tracking, and tyre scrubbing (wear). It also increases high-speed stability, reduces 'rearward amplification', and reduces the propensity to roll over in high-speed transient manoeuvres. The design of a novel experimental heavy goods vehicle with three independent hydraulically actuated steering axles is presented. The path-following controller is tested on the experimental vehicle, at low and high speeds. The field test results are compared with vehicle simulations and found to agree well. The benefits of this steering control approach are quantified. In a low-speed 'roundabout' manoeuvre, low-speed off-tracking was reduced by 73 per cent, from 4.25 m for a conventional vehicle to 1.15 m for the experimental vehicle; swept-path width was reduced by 2 m (28 per cent); peak scrubbing tyre forces were reduced by 83 per cent; and entry tail-swing was eliminated. In an 80 km/h lane-change manoeuvre, peak path error for the experimental vehicle was 33 per cent less than for the conventional vehicle, and rearward amplification of the trailer was 35 per cent less. Increasing the bandwidth of the steering actuators improved the high-speed dynamic performance of the vehicle, but at the expense of increased oil flow.

The infinite HMM for unsupervised PoS tagging

We extend previous work on fully unsupervised part-of-speech tagging. Using a non-parametric version of the HMM, called the infinite HMM (iHMM), we address the problem of choosing the number of hidden states in unsupervised Markov models for PoS tagging. We experiment with two non-parametric priors, the Dirichlet and Pitman-Yor processes, on the Wall Street Journal dataset using a parallelized implementation of an iHMM inference algorithm. We evaluate the results with a variety of clustering evaluation metrics and achieve equivalent or better performances than previously reported. Building on this promising result we evaluate the output of the unsupervised PoS tagger as a direct replacement for the output of a fully supervised PoS tagger for the task of shallow parsing and compare the two evaluations. © 2009 ACL and AFNLP.

Spatio-temporal clustering of probabilistic region trajectories

We propose a novel model for the spatio-temporal clustering of trajectories based on motion, which applies to challenging street-view video sequences of pedestrians captured by a mobile camera. A key contribution of our work is the introduction of novel probabilistic region trajectories, motivated by the non-repeatability of segmentation of frames in a video sequence. Hierarchical image segments are obtained by using a state-of-the-art hierarchical segmentation algorithm, and connected from adjacent frames in a directed acyclic graph. The region trajectories and measures of confidence are extracted from this graph using a dynamic programming-based optimisation. Our second main contribution is a Bayesian framework with a twofold goal: to learn the optimal, in a maximum likelihood sense, Random Forests classifier of motion patterns based on video features, and construct a unique graph from region trajectories of different frames, lengths and hierarchical levels. Finally, we demonstrate the use of Isomap for effective spatio-temporal clustering of the region trajectories of pedestrians. We support our claims with experimental results on new and existing challenging video sequences. © 2011 IEEE.