BMW and Daimler have announced that their joint mobility company will be headquartered in Berlin. The two companies have already filed documents for a joint venture with the European Commission. According to the two companies, Berlin was chosen due to its rich history of active usage of mobility services, as well as, due to its size, public transportation, and bicycling tradition.
The two automakers will each have 50% in the new joint venture but will continue competition in car production. According to BMW, the new company will cover five categories: multimodal and on-demand mobility (moovel and ReachNow), car-sharing (car2go and DriveNow), ride-hailing (myTaxi, Chauffeur Prive, Clever Taxi and Beat), parking (ParkNow and Parkmobile Group), and charging (ChargeNow and Digital Charging Solutions).
Uber in talks to acquire Deliveroo
Ride-hailing giant Uber is in talks to acquire food delivery company Deliveroo, according to Bloomberg. The terms of the negotiation are not known, however it is assumed that the price will be well above of the London-based Deliveroo’s current valuation — $2 billion. However, according to the reports, talks are fragile as Deliveroo and its investors prefer to be independent.
Deliveroo started its operations in London in 2013 and is now present in 12 countries.
nuTonomy releases dataset with 1.4 million images
Self-driving car company nuTonomy and Scale have released a dataset for autonomous vehicles with 1.4 million images. The product is called nuScenes and claims to surpass in size and accuracy public datasets like Kitti, ApolloScape (Baidu) and Udacity car library.
The dataset includes 1000 scenes with 1.4 million images, 400,000 lidar sweeps and 1.1 million 3 dimensional bounding boxes (objects detected with RGB cameras, radar and lidar). All images have been labeled through Scale’s Sensor Fusion Annotation API.
More and more companies are releasing self-driving car datasets, along with big companies, smaller ones like Flir Systems (released 10,000 photos), Mapillary (25,000 street-level images) and University of California Berkeley (100,000 video sequences) have all made their data-sets available to researchers.