Lyft shifts to open source

Lyft released an open source dataset of autonomous cars on July 23rd 2019. These data were collected on the company vehicles’ fleet by seven cameras and three Lidar per autonomous car. In addition, there are 55,000 manually annotated 3D images. In addition to this release, the company is announcing the organization of “a contest on this data set. The theme: detection of 3D objects on semantic maps”. This contest aims at developing machine learning models based on shared data. There is a $ 25,000 bonus to gain.

Which impact on the development of the autonomous vehicle?

Lyft intends to change the research approach on autonomous vehicle by removing the barrier of entry ticket: the access to data. The expected impacts are time saving in project implementation, a new generation of players, an expertise-driven market. The Territories, which are the main playgrounds for autonomous experiments, could play a more active role in their deployment.

We see here a handover of knowledge and skills which would announce a new organization of the market in both its structure and its business model. New models could emerge.

Which opportunity for the end user?

Many studies on the acceptability of this technology are being carried out. However, so far, the end user has not been sufficiently involved in the process.

Sharing information in open access could enable to involve the end user in the co-creation of an adapted offer, which will therefore limit the effort of acceptability.

The creation of a platform which would centralize all progress and news in this field and make them available to all represents an opportunity that has yet to be explored. It is essential to join this initiative and make the link between all involved actors. So far, the cooperation at national scale remains modest.

To conclude, the challenge is about encouraging cooperation to accelerate research and delivering the promise of the first functional autonomous vehicles by 2022.