Speed Daemon: Experience-Based Mobile Robot Speed Scheduling

Ostafew, Chris J.; Schoellig, Angela P.; Barfoot, Timothy D.; Collier, Jack

doi:10.1109/crv.2014.16

Cited by 16 publications

(10 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We implemented an automated speed scheduler (Ostafew et al., ) to demonstrate the LB‐NMPC algorithm's ability to interpolate and extrapolate from learned experiences. The algorithm uses experience from previous trials to schedule speeds that minimize travel time while ensuring reliable localization, low path‐tracking errors, and realizable control inputs.…”

Section: Methodsmentioning

confidence: 99%

See 1 more Smart Citation

Learning-based Nonlinear Model Predictive Control to Improve Vision-based Mobile Robot Path Tracking

et al. 2015

Self Cite

View full text Add to dashboard Cite

This paper presents a Learning‐based Nonlinear Model Predictive Control (LB‐NMPC) algorithm to achieve high‐performance path tracking in challenging off‐road terrain through learning. The LB‐NMPC algorithm uses a simple a priori vehicle model and a learned disturbance model. Disturbances are modeled as a Gaussian process (GP) as a function of system state, input, and other relevant variables. The GP is updated based on experience collected during previous trials. Localization for the controller is provided by an onboard, vision‐based mapping and navigation system enabling operation in large‐scale, GPS‐denied environments. The paper presents experimental results including over 3 km of travel by three significantly different robot platforms with masses ranging from 50 to 600 kg and at speeds ranging from 0.35 to 1.2 m/s (associated video at http://tiny.cc/RoverLearnsDisturbances). Planned speeds are generated by a novel experience‐based speed scheduler that balances overall travel time, path‐tracking errors, and localization reliability. The results show that the controller can start from a generic a priori vehicle model and subsequently learn to reduce vehicle‐ and trajectory‐specific path‐tracking errors based on experience.

show abstract

Section: Methodsmentioning

confidence: 99%

“…In this system, the LB‐NMPC algorithm is also shown to interpolate and extrapolate from experience. Preliminary results for the scheduler operating with a fixed feedback controller have been published in Ostafew, Collier, Schoellig, & Barfoot ().…”

Section: Introductionmentioning

confidence: 99%

Learning-based Nonlinear Model Predictive Control to Improve Vision-based Mobile Robot Path Tracking

et al. 2015

Self Cite

View full text Add to dashboard Cite

show abstract

“…We implemented an automated speed scheduler (Ostafew et al, 2014a) to demonstrate the LB-NMPC algorithm's ability to interpolate and extrapolate from learned experiences. The algorithm uses experience from previous trials to schedule speeds that minimize travel time while ensuring reliable localization, low path-tracking errors, and realizable control inputs.…”

Section: Automated Speed Schedulermentioning

confidence: 99%

“…In this system, the LB-NMPC algorithm is also shown to interpolate and extrapolate from experience. Preliminary results for the scheduler operating with a fixed feedback controller have been published in Ostafew et al (2014a).…”

Section: Introductionmentioning

confidence: 99%

Learning-based nonlinear model predictive control to improve vision-based mobile robot path-tracking in challenging outdoor environments

Ostafew

Schoellig

Barfoot

2014

2014 IEEE International Conference on Robotics and Automation (ICRA)

Self Cite

View full text Add to dashboard Cite

This paper presents a Learning-based Nonlinear Model Predictive Control (LB-NMPC) algorithm to achieve high-performance path tracking in challenging off-road terrain through learning. The LB-NMPC algorithm uses a simple a priori vehicle model and a learned disturbance model. Disturbances are modelled as a Gaussian Process (GP) as a function of system state, input, and other relevant variables. The GP is updated based on experience collected during previous trials. Localization for the controller is provided by an on-board, vision-based mapping and navigation system enabling operation in large-scale, GPS-denied environments. The paper presents experimental results including over 3 km of travel by three significantly different robot platforms with masses ranging from 50 kg to 600 kg and at speeds ranging from 0.35 m/s to 1.2 m/s. 1 Planned speeds are generated by a novel experience-based speed scheduler that balances overall travel time, path-tracking errors, and localization reliability. The results show that the controller can start from a generic a priori vehicle model and subsequently learn to reduce vehicle-and trajectory-specific path-tracking errors based on experience. 1 Associated video at http://tiny.cc/RoverLearnsDisturbances DMRV ROC6 Clearpath Husky Mass 600 kg 150 kg 50 kg Size (LW) 2 m x 1.5 m 1.5 m x 0.5 m 0.9 m x 0.6 m Steering Ackermann Steering Skid Steering Skid Steering Figure 1: Robots used to demonstrate the effectiveness of the learning controller. Despite significant differences in robot mass, wheel base, kinematics, and actuator designs, the algorithm uses the same nominal model for all three robots and learns disturbances over trials in order to accurately track desired paths.

show abstract

“…The goal of the task scheduler is to find an optimal assignment of a list of robotic tasks such that the schedule length or duration is minimised and in the same time the precedence constraints are preserved [1]. The conventional approach to schedule robotic tasks is to use a greedy algorithm in which the robot picks the nearest task or with the lowest cost function [2][3][4]. This strategy has the advantage of low computational time required but produces a non-optimal path [5].…”

Section: Introductionmentioning

confidence: 99%

A genetic algorithm based task scheduling system for logistics service robots

Harun¹,

Ibrahim²

2019

Bulletin EEI

View full text Add to dashboard Cite

The demand for autonomous logistics service robots requires an efficient task scheduling system in order to optimise cost and time for the robot to complete its tasks. This paper presents a Genetic algorithm (GA) based task scheduling system for a ground mobile robot that is able to find a global near-optimal travelling path to complete a logistics task of pick-and-deliver items at various locations. In this study, the chromosome representation and the fitness function of GA is carefully designed to cater for a single load logistics robotic task. Two variants of GA crossover are adopted to enhance the performance of the proposed algorithm. The performance of the scheduling is compared and analysed between the proposed GA algorithms and a conventional greedy algorithm in a virtual map and a real map environments that turns out the proposed GA algorithms outperform the greedy algorithm by 40% to 80% improvement.

show abstract

Speed Daemon: Experience-Based Mobile Robot Speed Scheduling

Cited by 16 publications

References 15 publications

Learning-based Nonlinear Model Predictive Control to Improve Vision-based Mobile Robot Path Tracking

Learning-based Nonlinear Model Predictive Control to Improve Vision-based Mobile Robot Path Tracking

Learning-based nonlinear model predictive control to improve vision-based mobile robot path-tracking in challenging outdoor environments

A genetic algorithm based task scheduling system for logistics service robots

Contact Info

Product

Resources

About