The self-driving truck startup selected the Matrix HPC storage system to provide flash-based parallel file storage capabilities.
WekalO announced Tuesday that TuSimple, a self-driving truck startup, had selected its Matrix HPC storage system to provide flash-based parallel file storage capabilities to accelerate its deep neural network (DNN) machine learning training.
TuSimple is in the process of developing a Level 4 autonomous truck-driving solution for the dock-to-dock delivery of commercial goods.
“WekaIO Matrix was the clear choice for our on-premises DNN training in the U.S.,” TuSimple Co-founder and CTO Dr. Xiaodi Hou said. “It was understood from the outset that a standard network-attached storage (NAS) solution would not be able to scale to the extent we would need it to, and apart from Matrix being the most performant of all the parallel file systems we evaluated, we really liked the fact that it is hardware-independent, allowing us better control over our infrastructure costs.
“We are also taking full advantage of WekaIO’s object storage capability, which is much more economical than an all-flash system, and allows us to efficiently scale our data catalog in a single namespace,” Hou added.
“We don’t rely on LiDAR as our primary sensor, we do a lot of camera-based analysis,” Hou said. “The data sets that train our AI models are comprised of millions of image files, which need to be read at high bandwidth. Matrix provides the low latency, high bandwidth we need to meet our data ingest demands.”
Chuck Price, vice president of product at TuSimple, told American Shipper in April that TuSimple had chosen to use cameras to achieve the 360-degree view because it is not only cheaper than lidar, but it can sense objects at a farther distance. Cameras require a lot more software technology to make them work, which has caused many autonomous vehicle startups to turn to lidar, Price said.
TuSimple taps WekaIO for machine training
