Robotics: Science and Systems XI 2015
DOI: 10.15607/rss.2015.xi.012
|View full text |Cite
|
Sign up to set email alerts
|

DeepMPC: Learning Deep Latent Features for Model Predictive Control

Abstract: Designing controllers for tasks with complex nonlinear dynamics is extremely challenging, time-consuming, and in many cases, infeasible. This difficulty is exacerbated in tasks such as robotic food-cutting, in which dynamics might vary both with environmental properties, such as material and tool class, and with time while acting. In this work, we present DeepMPC, an online real-time model-predictive control approach designed to handle such difficult tasks. Rather than hand-design a dynamics model for the task… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
1
1
1
1

Citation Types

0
218
0

Year Published

2017
2017
2021
2021

Publication Types

Select...
4
3
2

Relationship

0
9

Authors

Journals

citations
Cited by 274 publications
(218 citation statements)
references
References 28 publications
0
218
0
Order By: Relevance
“…Previous work on robotic cutting has focused on modelpredictive control (MPC) to generalize between different types of food items. Lenz et al [3] proposed learning a deep network to model the cutting interaction's dynamics for their DeepMPC approach. This network continually estimated the latent material properties throughout the cutting process using interactive perception and a recurrent network structure.…”
Section: Related Workmentioning
confidence: 99%
“…Previous work on robotic cutting has focused on modelpredictive control (MPC) to generalize between different types of food items. Lenz et al [3] proposed learning a deep network to model the cutting interaction's dynamics for their DeepMPC approach. This network continually estimated the latent material properties throughout the cutting process using interactive perception and a recurrent network structure.…”
Section: Related Workmentioning
confidence: 99%
“…Recent papers have explored model-predictive control with deep networks [21], [22], [23], [24], [25]. These approaches learn an abstract-state transition function, not an explicit model of the environment [26], [27].…”
Section: Control With a Learned Simulatormentioning
confidence: 99%
“…Lenz, Knepper, and Saxena [122] modeled robotic food cutting with a knife. This includes difficult-to-model effects such as friction, deformation, and hysteresis.…”
Section: Examples In Recent Researchmentioning
confidence: 99%
“…This multi-step approach sped both training and classification at runtime. Lenz et al [122] employed a two-stage network design for grasp detection. The first DNN had relatively few parameters.…”
Section: Current Shortcomings Of Dnns For Roboticsmentioning
confidence: 99%