AImotive Looks to Make an Affordable Alternative to LiDAR

Home > News > Content

【Summary】Software company AImotive is looking for a way to make autonomous cars cheaper and easier to manufacture with a LiDAR alternative.

Original   Vineeth Joel Patel  ·  May 20, 2017 8:05 AM PT
author: Vineeth Joel Patel   

The automotive industry is seeing an influx of technology companies and automakers that are working on artificial intelligence for automobiles. The idea behind giving cars artificial intelligence software is that the vehicles will be able to see the road for themselves, learning what real-life objects look like and how to avoid them. 

Now, a new software company, AImotive, is looking to shake the industry up by attempting to create an affordable AI system that doesn't use a laser-based LiDAR system and by creating the necessary hardware, as well. 

How AImotive Plans To Replace LiDAR

As Engadget reports, AImotive is working on creating a more affordable autonomous system by passing LiDAR sensors in favor of numerous low-cost cameras. LiDAR is widely used across the entire industry and through the use of lasers, allows a vehicle to map out its environment. 

Velodyne, a California-based manufacturer of LiDAR sensors, introduced a solid state LiDAR sensor for autonomous vehicles named "Velarray." The new sensor costs approximately $80,000 each, which raises the price for self-driving cars. AImotive, as Engadget claims, is looking to ditch Velodyne's widely-used LiDAR sensor for eight to 12 "off-the-shelf" cameras that cost $12 a piece. 

A report by The Drive, claims that AImotive's system works in a similar fashion to a human eye. Each pair of cameras faces a different location to sense depth and also provide a 360-degree look of the car's surroundings, claims the outlet. While not as high-tech as a LiDAR system, the cameras are capable of doing the same thing by detecting where road lines are, objects on the road, and other driving-related items. 

What Else Is AImotive Working On?

To test its hardware and software systems, AImotive upgraded a Toyota Prius. To make sense of all of the data that's coming in from the cameras, the software company has gone with a platform from NVIDIA, reports Engadget. While the report didn't state what type of platform the car is using, AImotive is looking for a cheaper alternative route by building its own computer, claims the outlet. 

Building its own hardware to retrieve and comprehend the data will be an uphill battle, but NVIDIA's system uses approximately 1,000 watts of power, which would put a strain on electric cars. AImotive also plans to sell its computer system to other companies and automakers. 

As The Drive reports, AImotive's computer system will be based on custom Application-Specific Integrated Circuits (ASICs) that will reduce the power consumption from 1,000 to 100 watts The cost will go down as well, as NVIDIA Titan X's graphic cards go for $1,000 apiece, to something more reasonable at $100 or less, reports the outlet. 

Engadget reports that AImotive wants self-driving cars to act more like a human. The company, as the outlet reports, believes that radar, or more specifically LiDAR, should be used as a backup, while a vision-based system should be the primary method of viewing the world. 

AImotive definitely has an uphill battle when it comes to competing against the likes of NVIDIA, Intel, and Velodyne. But competition will help lead the way to more affordable autonomous vehicles that are easier to design. 

via: The Drive, Engadget


Prev                  Next
Writer's other posts
Comments
Comments:
    . Related Content