• Corpus ID: 195707955

Incremental Learning for Motion Prediction of Pedestrians and Vehicles

@inproceedings{Govea2010IncrementalLF,
  title={Incremental Learning for Motion Prediction of Pedestrians and Vehicles},
  author={Alejandro Dizan Vasquez Govea},
  year={2010}
}
Abstract The main subject of this thesis is motion prediction. The problem is approached from the hypothesis that the dynamic and kinematic properties of objects such as pedestrian and vehicles do not suffice to predict their motion in the long term. Instead, the work presented here, in scribes itself in a new family of approaches which assume that, in a given environment, objects do not move at random, but engage in “typical motion patterns”, which may be learned and then used to predict motion… 

Incremental Learning of Statistical Motion Patterns With Growing Hidden Markov Models

TLDR
This work presents an approach where motion patterns can be learned incrementally and in parallel with prediction, based on a novel extension to hidden Markov models (HMMs) - called growing hidden MarkOV models - which gives the ability to incrementally learn both the parameters and the structure of the model.

Modeling and learning behaviors

TLDR
This section discusses Growing Hidden Markov Models, an approach which is able to perform lifelong learning, continuously updating its knowledge as more data are available, and in experimental evaluation against two other state-of-the-art approaches, the presented approach consistently outperforms them regarding both prediction accuracy and model parsimony.

Contribution to Perception for Intelligent Vehicles

TLDR
A model-based approach is introduced to interpret the laser measurement sequence over a sliding window of time by hypotheses of moving object trajectories by exploring the data-driven Markov chain Monte Carlo technique.

Bayesian environment representation, prediction, and criticality assessment for driver assistance systems

TLDR
Parametric Free Space maps are introduced, which compactly represent the vehicle environment in form of relevant, drivable free space, and a maneuver-based, long-term trajectory prediction and criticality assessment system is introduced.

Data-based reachability analysis for movement prediction of pedestrians and motion planning

TLDR
A state of the art motion planning approach with Mixed-Integer Linear optimization (MILP) is used for the trajectory planning of the vehicle and prior knowledge can be incorporated in the reachability analysis, and the computional load is scalable.

Safe Motion Planning for Autonomous Driving

TLDR
A novel motion planning system is described, which translates high-level navigation goals into low-level actions for controlling a vehicle, and is able to compute a safe and efficient trajectory in a dynamic environment while staying within its lane and avoiding obstacles.

Analysis of Dynamic Scenes: Application to Driving Assistance

TLDR
This thesis proposes a complete probabilistic model motion at the trajectory level based the Gaussian Process (GP), whose advantage over current methods is that it is able to express future motion independently of state space discretization.

Risk based motion planning and navigation in uncertain dynamic environment

TLDR
A new concept to integrate a probabilist collision risk function linking planning and navigation methods with the perception and the prediction of the dynamic environments, and shows the performance for a robotic wheelchair in a simulated environment among multiple dynamic obstacles.

Predicting Next Points of Interests Based on a Markov Model

TLDR
A novel method based on Markov chain for prediction, which uses GEPETO variant algorithm to cluster for POIs to solve the former algorithm without considering the temporal factor and presents Mobility Markov Chain (MMC) model which exploits 3 previous states to infer the future location.

Optimal motion planning with reachable sets of vulnerable road users

TLDR
A new jerk constrained model for the pedestrian is introduced and combined with a Gaussian Process prediction model to predict future trajectory with absolute certainty and it can be proven for all kind of pedestrians.

References

SHOWING 1-10 OF 103 REFERENCES

Modeling and Recognition of Human Actions Using a Stochastic Approach

TLDR
A self-learning prototype system for the real-time detection of unusual motion patterns and motion recognition based on the same method, the extended Condensation algorithm, used for the object tracking.

Learning Motion Patterns of People for Compliant Robot Motion

TLDR
A technique for learning collections of trajectories that characterize typical motion patterns of persons and how to incorporate the probabilistic belief about the potential trajectories of persons into the path planning process of a mobile robot is proposed.

Predictive autonomous robot navigation

TLDR
This paper considers the problem of a robot navigating in a crowded or congested environment and proposes a hierarchical representation of POMDPs to attempt to predict the motion trajectory of humans and obstacles.

Learning motion patterns of persons for mobile service robots

TLDR
An algorithm that learns collections of typical trajectories that characterize a person's motion patterns, related to specific locations that they might be interested in approaching and specific trajectories they might follow in doing so is proposed.

Motion prediction for moving objects: a statistical approach

  • D. VasquezThierry Fraichard
  • Computer Science
    IEEE International Conference on Robotics and Automation, 2004. Proceedings. ICRA '04. 2004
  • 2004
TLDR
A technique to obtain long term estimates of the motion of a moving object in a structured environment by observing the environment and clustering the observed trajectories using any pairwise clustering algorithm.

Moving object prediction for off-road autonomous navigation

TLDR
A combined probabilistic object classification and estimation theoretic framework to predict the future location of moving objects, along with an associated uncertainty measure is outlined.

Learning Patterns of Activity Using Real-Time Tracking

TLDR
This paper focuses on motion tracking and shows how one can use observed motion to learn patterns of activity in a site and create a hierarchical binary-tree classification of the representations within a sequence.

Camera-based observation of obstacle motions to derive statistical data for mobile robot motion planning

  • E. KruseF. Wahl
  • Computer Science
    Proceedings. 1998 IEEE International Conference on Robotics and Automation (Cat. No.98CH36146)
  • 1998
TLDR
This paper presents concepts and results of an experimental system using cameras mounted at the ceiling, the workspace is observed, and the image data is processed and transformed into a compact statistical representation of motion patterns that yields efficient paths which are adapted to obstacle behavior.

Motion prediction of moving objects based on autoregressive model

TLDR
A framework for predicting future positions and orientation of moving obstacles in a time-varying environment using autoregressive model (ARM) with conditional maximum likelihood estimate of the model parameters is described.
...