Future vehicle systems for active pedestrian safety will not only require a high recognition performance but also an accurate analysis of the developing traffic situation. In this paper, we present a study on pedestrian path prediction and action classification at short subsecond time intervals. We consider four representative approaches: two novel approaches (based on Gaussian process dynamical models and probabilistic hierarchical trajectory matching) that use augmented features derived from dense optical flow and two approaches as baseline that use positional information only (a Kalman filter and its extension to interacting multiple models). In experiments using stereo vision data obtained from a vehicle, we investigate the accuracy of path prediction and action classification at various time horizons, the effect of various errors (image localization, vehicle egomotion estimation), and the benefit of the proposed approaches. The scenario of interest is that of a crossing pedestrian, who might stop or continue walking at the road curbside. Results indicate similar performance of the four approaches on walking motion, with near-linear dynamics. During stopping, however, the two newly proposed approaches, with nonlinear and/or higher order models and augmented motion features, achieve a more accurate position prediction of 10-50 cm at a time horizon of 0-0.77 s around the stopping event.

/pdf/will-the-pedestrian-cross-a-study-on-pedestrian-path-3gjg454anu.pdf

Will the Pedestrian Cross? A Study on Pedestrian Path Prediction

A typical sensor data processing sequence uses a detection algorithm prior to tracking to extract point-measurements from the observed sensor data. Track-before-detect (TkBD) is a paradigm which combines target detection and estimation by removing the detection algorithm and supplying the sensor data directly to the tracker. Various different approaches exist for tackling the TkBD problem. This paper compares the ability of several different approaches to detect low amplitude targets. The following algorithms are considered in this comparison: Bayesian estimation over a discrete grid, Dynamic Programming, Particle Filtering methods, and the Histogram Probabilistic Multi-Hypothesis Tracker. Algorithms are compared on the basis of detection performance and computation resource requirements.

/pdf/a-comparison-of-detection-performance-for-several-track-4g0umeuwaa.pdf

A comparison of detection performance for several Track-Before-Detect algorithms

Bernoulli filters are a class of exact Bayesian filters for non-linear/non-Gaussian recursive estimation of dynamic systems, recently emerged from the random set theoretical framework. The common feature of Bernoulli filters is that they are designed for stochastic dynamic systems which randomly switch on and off. The applications are primarily in target tracking, where the switching process models target appearance or disappearance from the surveillance volume. The concept, however, is applicable to a range of dynamic phenomena, such as epidemics, pollution, social trends, etc. Bernoulli filters in general have no analytic solution and are implemented as particle filters or Gaussian sum filters. This tutorial paper reviews the theory of Bernoulli filters as well as their implementation for different measurement models. The theory is backed up by applications in sensor networks, bearings-only tracking, passive radar/sonar surveillance, visual tracking, monitoring/prediction of an epidemic and tracking using natural language statements. More advanced topics of smoothing, multi-target detection/tracking, parameter estimation and sensor control are briefly reviewed with pointers for further reading.

A Tutorial on Bernoulli Filters: Theory, Implementation and Applications

The paper deals with a radar ‘track before detect’ application in a multitarget setting ‘Track before detect’ is a method to track weak objects on the basis of raw radar measurements, eg the reflected power of the target plus noise In classical target tracking, the tracking process is performed on the basis of pre-processed measurements that are constructed from the original measurement data every time step In this way no integration over time takes place and information is lost The authors give details of a modelling setup and a particle filter based algorithm to deal with a ‘multiple target track before detect’ situation Using simulations it is shown that with this method, it is possible to track multiple, closely spaced, weak targets

Multitarget particle filter track before detect application

We propose a filtering framework for multitarget tracking that is based on the probability hypothesis density (PHD) filter and data association using graph matching. This framework can be combined with any object detectors that generate positional and dimensional information of objects of interest. The PHD filter compensates for missing detections and removes noise and clutter. Moreover, this filter reduces the growth in complexity with the number of targets from exponential to linear by propagating the first-order moment of the multitarget posterior, instead of the full posterior. In order to account for the nature of the PHD propagation, we propose a novel particle resampling strategy and we adapt dynamic and observation models to cope with varying object scales. The proposed resampling strategy allows us to use the PHD filter when a priori knowledge of the scene is not available. Moreover, the dynamic and observation models are not limited to the PHD filter and can be applied to any Bayesian tracker that can handle state-dependent variances. Extensive experimental results on a large video surveillance dataset using a standard evaluation protocol show that the proposed filtering framework improves the accuracy of the tracker, especially in cluttered scenes.

/pdf/efficient-multitarget-visual-tracking-using-random-finite-39wuz9zf4h.pdf

Efficient Multitarget Visual Tracking Using Random Finite Sets

A new method for multiple model particle filtering for Markovian switching systems is presented. This new method is a combination of the interacting multiple model (IMM) filter and a (regularised) particle filter. The mixing and interaction is similar to that in a conventional IMM filter. However, in every mode a regularised particle filter is running. The regularised particle filter probability density is a mixture of Gaussian probability densities. The proposed method is able to deal with nonlinearities and non-Gaussian noise. Furthermore, the new method keeps a fixed number of particles in each mode, and therefore it does not suffer from the potential drawbacks of existing multiple model particle filters for Markovian switching systems.

Interacting multiple model particle filter

We present a new method to perform detection and tracking of a possible target in noise. We perform tracking not on the basis of the standard measurements but on the raw radar video data. Detection then is based upon the a posteriori information, i.e., the probability density of the state given these past measurements (in this case video data). This way of data processing/tracking is also referred to as track before detect (TBD) for obvious reasons. An advantage of this method over classical tracking is that in this TBD approach the decision whether a target is present or not is based on integrated and kinematically correlated energy. This method is better suited for tracking weak targets in noise than the classical method. As this problem statement leads to a nonlinear non-Gaussian filtering problem classical filtering methods (e.g. Kalman filtering) will result in poor performance. A particle filter is used to deal with the nonlinearities and the non-Gaussian nature of the noise. The same particle filter output is also used to perform detection based on a likelihood ratio test.

Particle filter based detection for tracking

We consider the smoothing problem for a general state space system using sequential Monte Carlo (SMC) methods. The marginal smoother is assumed to be available in the form of weighted random particles from the SMC output. New algorithms are developed to extract the smoothed marginal maximum a posteriori (MAP) estimate of the state from the existing marginal particle smoother. Our method does not need any kernel fitting to obtain the posterior density from the particle smoother. The proposed estimator is then successfully applied to find the unknown initial state of a dynamical system and to address the issue of parameter estimation problem in state space models.

/pdf/particle-based-smoothed-marginal-map-estimation-for-general-2l6gle0u38.pdf

J.N. Driessen

Papers

Interacting multiple model particle filter

Multitarget particle filter track before detect application

Particle filter based detection for tracking

Particle Based Smoothed Marginal MAP Estimation for General State Space Models