Difference between revisions of "RollE affordable modular autonomous vehicle development platform"

From Communauté de la Fabrique des Mobilites
Line 6: Line 6:
  
 
This [https://ieeexplore.ieee.org/document/8506757 paper] describes the design of RollE, an affordable modular autonomous vehicle development platform. It is capable of driving via remote control for data collection and also capable of autonomous drivingusing a convolutional neural network. This system is aimed at providing students and researchers with an affordable autonomous vehicle to develop and test self-driving car technology.
 
This [https://ieeexplore.ieee.org/document/8506757 paper] describes the design of RollE, an affordable modular autonomous vehicle development platform. It is capable of driving via remote control for data collection and also capable of autonomous drivingusing a convolutional neural network. This system is aimed at providing students and researchers with an affordable autonomous vehicle to develop and test self-driving car technology.
 +
 +
{{#widget:SlideShare|id=131894992|width=500|height=500|alignment=center}}
 
|Tags=autonomous car, raspberry, DIY
 
|Tags=autonomous car, raspberry, DIY
 
|commonscategory=Logiciel, Données, Connaissance, Matériel
 
|commonscategory=Logiciel, Données, Connaissance, Matériel
Line 14: Line 16:
 
|chat=https://chat.fabmob.io/channel/navette_autonome
 
|chat=https://chat.fabmob.io/channel/navette_autonome
 
}}
 
}}
 +
Road accidents are estimated to be the ninth leading cause of death across all age groups globally. The annual estimated global tally of deaths as a result of road accidents hovers around 1.25 million people. These accidents are mostly due to preventable human driver error and autonomous vehicles provide a prospective solution to this problem. Interest in the potential of autonomous vehicles has grown substantially in the past four years. As of June 2017, the research institution Brookings estimated the total investment in research and development of autonomous vehicles by industry leaders to have grown from under $1 billion in late 2014 to about $80 billion.
 +
Financial constraints are a major roadblock to self-driving research as typical autonomous vehicle projects require considerable resources. For instance, Stanford university’s autonomous vehicle Stanley required dedicated funding from sponsors such as VW, Redbull and Intel, among others. Equipment typically used in these autonomous cars such as Velodyne scanners and ring laser gyroscopes can cost as much as $85,000 and $20,000 respectively.
 +
There exists a plethora of commercial and open source robot programming simulation software such as Gazebo designed to abstract hardware and provide sophisticated tools for single and multi-robot programming. These options seem viable for low cost experimentation involving autonomous robots, and, in fact, a majority of algorithms can be tested in simulated environments. However, the practical usefulness of these simulated environments in validating algorithms for real life probabilistic tasks can be diminished due to the uncertainties, constraints and challenges introduced by the real world. This makes it imperative that self-driving research be conducted on physical hardware subject to these constraints and challenges and necessitates the existence of an affordable physical platform for cost efficient self-driving research.
 +
This workshop introduces and describes the design of RollE, a novel affordable modular scaled-down autonomous vehicle platform designed to reduce the barrier to entry into self- driving research, in terms of cost of equipment. This platform aims to accelerate self-driving research by providing students and researchers with a low-cost autonomous vehicle to test ideas and build self-driving technology using machine learning techniques similar to those used in the industry.

Revision as of 09:20, 15 February 2019


Description en une ligne : RollE affordable modular autonomous vehicle development platform

Description : RollE is an affordable modular autonomous vehicle development platform developed to reduce the barrier to entry for self-driving research. It is capable of driving via remote control for data collection and autonomous driving using a convolutional neural network. RollE demonstrates that an affordable autonomous vehicle built with hobby grade electronics can effectively be used to implement self-driving ideas using scalable technology such as machine learning. Read Benedict Quartey story

This paper describes the design of RollE, an affordable modular autonomous vehicle development platform. It is capable of driving via remote control for data collection and also capable of autonomous drivingusing a convolutional neural network. This system is aimed at providing students and researchers with an affordable autonomous vehicle to develop and test self-driving car technology.

Image :

Ashesi BenedictQuartey inline1.jpg

Site internet : https://www.youtube.com/watch?v=1iLejcGQvJw

Organisations intéressées pour contribuer ou qui contribuent déjà :

Organisations utilisatrice ou intéressée par utiliser la ressource :

Contributeurs :

Tags : autonomous car, raspberry, DIY

Catégories : Logiciel, Données, Connaissance, Matériel

Thème : Open HardWare, Navettes autonomes

Référent : ROESCH Nicolas

Défi auquel répond la ressource : Abaisser les barrières pour innover sur le véhicule

Personnes clés à solliciter :

Autre commun proche :

Communauté d'intérêt :

Type : Commun

Type de licence ?

Conditions Générales d'Utilisation (CGU) :

Niveau de développement :

Espace d'échange synchrone (de type chat) : https://chat.fabmob.io/channel/navette_autonome

Espace d'échange synchrone (de type forum) : [1]

Lien vers l'outil de gestion des actions :

Lien vers l'outil de partage de fichiers :

Besoins :

Complément :

Prochaine Etape :



Autres informations :

Liste des Acteurs qui utilisent ou souhaitent utiliser ce Commun RollE affordable modular autonomous vehicle development platform : aucun pour le moment

Road accidents are estimated to be the ninth leading cause of death across all age groups globally. The annual estimated global tally of deaths as a result of road accidents hovers around 1.25 million people. These accidents are mostly due to preventable human driver error and autonomous vehicles provide a prospective solution to this problem. Interest in the potential of autonomous vehicles has grown substantially in the past four years. As of June 2017, the research institution Brookings estimated the total investment in research and development of autonomous vehicles by industry leaders to have grown from under $1 billion in late 2014 to about $80 billion. Financial constraints are a major roadblock to self-driving research as typical autonomous vehicle projects require considerable resources. For instance, Stanford university’s autonomous vehicle Stanley required dedicated funding from sponsors such as VW, Redbull and Intel, among others. Equipment typically used in these autonomous cars such as Velodyne scanners and ring laser gyroscopes can cost as much as $85,000 and $20,000 respectively. There exists a plethora of commercial and open source robot programming simulation software such as Gazebo designed to abstract hardware and provide sophisticated tools for single and multi-robot programming. These options seem viable for low cost experimentation involving autonomous robots, and, in fact, a majority of algorithms can be tested in simulated environments. However, the practical usefulness of these simulated environments in validating algorithms for real life probabilistic tasks can be diminished due to the uncertainties, constraints and challenges introduced by the real world. This makes it imperative that self-driving research be conducted on physical hardware subject to these constraints and challenges and necessitates the existence of an affordable physical platform for cost efficient self-driving research. This workshop introduces and describes the design of RollE, a novel affordable modular scaled-down autonomous vehicle platform designed to reduce the barrier to entry into self- driving research, in terms of cost of equipment. This platform aims to accelerate self-driving research by providing students and researchers with a low-cost autonomous vehicle to test ideas and build self-driving technology using machine learning techniques similar to those used in the industry.