Fully Autonomous Driving Back to Top
Analysis by: James Ingham and Jonathan Davenport
Description: Fully autonomous driving will employ the use of perception sensors tied to AI-based, decision-making algorithms on a compute platform to guide a vehicle through a road environment without human assistance. Fully autonomous driving will employ various sensing solutions and localization technologies, such as lidar, radar, cameras, global navigation satellite system (GNSS) and high-definition map data.
Sample Vendors: Aptiv; Aurora Innovation; Baidu (Apollo); Cruise; Intel (Mobileye); TuSimple; Uber; Waymo; Zoox
Range: Long (6 to 8 Years)
Fully autonomous driving is so challenging to achieve, it will not reach early majority adoption in the next eight years, despite significant industry investment. 1 The companies working on achieving vehicle autonomy have made tremendous progress, but remain challenged by a significant number of factors that impede scalable deployments. These include technical, infrastructure, policies, public opinion and finances. Weather conditions, the design of road networks and human driving styles vary significantly in different parts of the world, which makes training a single algorithm to cope with this wide range of eventualities challenging. At present, systems are being designed to work in geofenced operational design domains (ODDs), such as specific areas within cities or stretches of highway. In addition, the high-definition maps required by autonomous navigation systems are difficult to build and maintain over large geographic areas. Finally, autonomous systems have yet to demonstrate sufficient accuracy to handle the long tail of complex edge-case situations that exist in urban center environments.
Gartner, Inc. | G00786204
Page 34 of 48
This research note is restricted to the personal use of abhishek.sharma@fractal.ai.
Made with FlippingBook - PDF hosting