Home » 2016

Yearly Archives: 2016

Can a robot drive 100s of Kilometers with < 1 m Error?

A considerably difficult aspect of Simultaneous Localization and Mapping (SLAM) is the problem of uncertainty constrained long term point-to-point navigation where global loop closures to eliminate estimation biases may not be possible. In such scenarios, a prime concern is to control the rate of localization error growth. We have developed fundamental results on the underlying problem of localization and a novel SLAM technique that allows sub-meter localization error for a > 100 km trajectory without loop closure and using only on-board sensors, i.e., no GPS.

Applications may include:

  1. Self-driving cars
  2. Precision farming
  3. Planetary rovers
  4. Unmanned Aerial Vehicles
  5. Autonomous Underwater Vehicles

To license this technology for commercial use or to learn more, please contact lab director Dr. Suman Chakravorty or Dr. Ismail Sheikh (smismail[at]tamu[dot]edu) at Texas A&M University Technology Commercialization, 800 Raymond Stotzer Parkway, Suite 2020, College Station, Texas 77845.

EDPLab Develops a Novel Technique For Feature-Based SLAM

The SLAM problem is known to have a special property that when robot orientation is known, estimating the history of robot poses and feature locations can be posed as a standard linear least squares problem. We have developed a SLAM framework that uses relative feature-to-feature measurements to exploit this structural property of SLAM. Relative feature measurements are used to pose a linear estimation problem for pose-to-pose orientation constraints. This is followed by solving an iterative non-linear on-manifold optimization problem to compute the maximum likelihood estimate for robot orientation given relative rotation constraints. Once the robot orientation is computed, we solve a linear problem for robot position and map estimation. Our approach reduces the computational burden of non-linear optimization by posing a smaller optimization problem as compared to standard graph-based methods for feature-based SLAM. By separating orientation estimation and formulating the robot and landmark position estimation as a linear least squares problem, no initial guess is required for the positions. Further, empirical results show our method avoids catastrophic failures that arise in existing methods due to using odometery as an initial guess for non-linear optimization, while its accuracy degrades gracefully as sensor noise is increased.

[Feel free to study the paper (pdf) submitted for review to ICRA 2017.]



Dan Yu Successfully Defends Her Ph.D.!

Dan Defense

Our heartiest congratulations to Dr. Yu on successfully defending! Dr. Yu’s research focused on model order reduction techniques for large scale systems, you may find her publications here. Dr. Yu will continue her research at the EDP Lab.

Decentralized State Estimation via a Hybrid of Consensus and Covariance intersection – Technical Report

This paper presents a new recursive information consensus filter for decentralized dynamic-state estimation. No structure is assumed about the topology of the network and local estimators are assumed to have access only to local information. The network need not be connected at all times. Consensus over priors which might become correlated is performed through Covariance Intersection (CI) and consensus over new information is
handled using weights based on a Metropolis Hastings Markov Chain. We establish bounds for estimation performance and show that our method produces unbiased conservative estimates that are better than CI. The performance of the proposed method is evaluated and compared with competing algorithms on an atmospheric dispersion problem.

Here is a a technical report which is an extended version of the IROS submission with extra proofs and content. TechnicalReport