Canada Spends Logo

Explorez notre analyse du budget 2026

Grant: Discovery Grants Program - Individual

Simon Fraser University

$36,000.00

Machine Learning for Agile Motor Skills

EF 2024-25

Résumé
Artificial intelligence has seen remarkable breakthroughs in recent years, leading to systems that can accurately identify objects in complex scenes, generate functional code from natural language descriptions, synthesize high-resolution photo-realistic images, and surpass the best human players in challenging games. While these systems are able to achieve incredible cognitive feats, the physical capabilities of artificial agents still fall well behind the athletic prowess demonstrated by humans and other animals. Humans are capable of performing awe-inspiring feats of agility in complex environments by drawing from a vast repertoire of diverse and sophisticated motor skills. This dynamism is in sharp contrast to the narrowly specialized and rigid behaviors commonly exhibited by artificial agents in both simulated and real-world domains. This proposal describes a plan to develop general and scalable techniques that allow agents to replicate the motor capabilities of humans and animals, with a focus on applications in computer graphics and robotics. Our work will start by exploring reinforcement learning and imitation learning techniques in simulation, with the goal of developing controllers that enable simulated agents to perform complex motor control tasks. As the capabilities of agents improve, providing users with the means of directing the behaviors of these agents will be vital for practical applications of these systems. Our work will leverage advances from natural language processing to ground language commands in physical skills to enable agents to effectively follow verbal commands. Once our techniques have been validated in simulation, we will proceed to apply them to develop controllers for physical robots. Our work will aim to develop robotic systems that can agilely navigate and interact with unstructured real-world environments, with minimal instrumentation and a-priori information of the environment. Developing computational models that are able to replicate these agile behaviors has the potential for far-reaching applications in a myriad of fields. Effective motor control models for virtual characters can benefit a large array of applications that require generation of human motions, such as films, games, biomechanics, architecture, interior design, autonomous driving, therapeutics, prostheses, and many more. Language-directed virtual characters can allow users to generate animations more easily without requiring complex animation tools. More capable motor controllers can enable robots to venture out of the structured environments of labs and factories and perform tasks in complex real-world environments. This mobility can greatly expand the application areas in which robotic platforms can be effectively utilized.
Programme
Discovery Grants Program - Individual
Emplacement
British Columbia, Canada
Type de bénéficiaire
Academia
Type
Grant
Source
NSERC's Awards Data

Tel que publié

Application id
RGPIN-2023-05521
Name nom
Peng, Xue Bin
Department département
Computing Science
Organization id
5
Institution établissement
Simon Fraser University
Province en
British Columbia
Province fr
Colombie-Britannique
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2024
Competition year année de concours
2023
Award amount
36000
Program id
RGPIN
Program name en
Discovery Grants Program - Individual
Program name fr
Programme de subventions à la découverte - individuelles
Group en
DISCOVERY RESEARCH
Group fr
RECHERCHE AXÉE SUR LA DÉCOUVERTE
Committee code
1507
Committee name en
Computer Science
Committee name fr
Sciences informatiques
Area of application code
801
Area of application group en
Information and communication services
Area of application group fr
Services d'information et de télécommunications
Area of application en
Computer software
Area of application fr
Logiciel
Research subject code
2707
Research subject group en
Information technology
Research subject group fr
Technologie de l'information
Research subject en
Graphics
Research subject fr
Infographie
Field of research list names en
Not available
Field of research list names fr
Non disponible
Socioeconomic objective list names en
Not available
Socioeconomic objective list names fr
Non disponible
Application title
Machine Learning for Agile Motor Skills
Keywords
reinforcement learning; motor control; computer animation; robotic locomotion; machine learning; computer graphics
Application summary
Artificial intelligence has seen remarkable breakthroughs in recent years, leading to systems that can accurately identify objects in complex scenes, generate functional code from natural language descriptions, synthesize high-resolution photo-realistic images, and surpass the best human players in challenging games. While these systems are able to achieve incredible cognitive feats, the physical capabilities of artificial agents still fall well behind the athletic prowess demonstrated by humans and other animals. Humans are capable of performing awe-inspiring feats of agility in complex environments by drawing from a vast repertoire of diverse and sophisticated motor skills. This dynamism is in sharp contrast to the narrowly specialized and rigid behaviors commonly exhibited by artificial agents in both simulated and real-world domains. This proposal describes a plan to develop general and scalable techniques that allow agents to replicate the motor capabilities of humans and animals, with a focus on applications in computer graphics and robotics. Our work will start by exploring reinforcement learning and imitation learning techniques in simulation, with the goal of developing controllers that enable simulated agents to perform complex motor control tasks. As the capabilities of agents improve, providing users with the means of directing the behaviors of these agents will be vital for practical applications of these systems. Our work will leverage advances from natural language processing to ground language commands in physical skills to enable agents to effectively follow verbal commands. Once our techniques have been validated in simulation, we will proceed to apply them to develop controllers for physical robots. Our work will aim to develop robotic systems that can agilely navigate and interact with unstructured real-world environments, with minimal instrumentation and a-priori information of the environment. Developing computational models that are able to replicate these agile behaviors has the potential for far-reaching applications in a myriad of fields. Effective motor control models for virtual characters can benefit a large array of applications that require generation of human motions, such as films, games, biomechanics, architecture, interior design, autonomous driving, therapeutics, prostheses, and many more. Language-directed virtual characters can allow users to generate animations more easily without requiring complex animation tools. More capable motor controllers can enable robots to venture out of the structured environments of labs and factories and perform tasks in complex real-world environments. This mobility can greatly expand the application areas in which robotic platforms can be effectively utilized.