This post was originally published on this site
Last month, a faulty CrowdStrike Update brought down hospitals, 911 call centers, and airports. This showed the impact of a defective update on critical infrastructure. Imagine if this update was intended for an autonomous vehicle, or a robot in a warehouse. The consequences of a bad upgrade could be even worse.
Trace Machina is an early-stage startup that aims to prevent such scenarios by using simulation software. This allows developers to test updates within a more realistic simulated setting. The company emerged out of stealth on Thursday and announced a $4.7million seed investment as well as an open-source tool called NativeLink.
Marcus Eagan, CEO and cofounder of the company, says that his company is developing an native Rust-based software system to test and validate autonomous systems such as self-driving vehicles and warehouse automation equipment.
Eagan told TechCrunch that “the way we solve this is by providing a NativeLink between developers and their autonomy vision”. NativeLink is the name of the company’s very first product.
“When developers move from working on web applications to working on robots it becomes evident that the existing developer’s toolkit with Docker and Kubernetes etc. is not sufficient. This is not enough. He said that engineers need to be able run experiments and test directly on local hardware.
NativeLink bridges this gap and provides engineers a staging environment which enables them run simulations on resource-constrained platforms like an embedded Nvidia graphics chip, which is difficult to source for self-driving vehicles and edge devices.
Eagan says companies previously had to build their own environments, which limited them to hyper scalers or well-funded self driving car companies like Google. He wanted to create a system as close as possible to the hardware, which he calls “being as close as the metal,” to make it available to any company.
“There are a lot people who have gone down this road, but none can run with direct access to hardware.” There was always this virtualized layer. This abstraction layer made it easier for these companies to build and iterate their systems. “We just had to pay a tax for being so close to the metal”, he said.
Eagan has worked at MongoDB and helped develop Atlas Vector Search, MongoDB’s first AI product. Nathan Bruer was his co-founder and worked at Google X – the company’s moonshot project center – as well as at Toyota Institute, where he helped build autonomous cars.
Eagan is Black and has dealt with racism throughout his career. He remains focused on growing his company. “I have dealt with racism, and I don’t mind. I’m so focused. Nobody can stop or dictate what I do. “I’m grateful for this freedom from that perspective because many people who look like myself don’t have it,” he said.
He has had to overcome many obstacles in his life, including racism. When he was a teenager, he was involved in a serious car accident that left him severely injured and unable to speak or walk. He was able recover, attend college, become an engineering student, and eventually start this startup.
The $4.7 million seed investment was led by Wellington Management, with participation from Samsung Next and Sequoia Capital Scout Fund. Green Bay Ventures, Verissimo Ventures as well as several prominent industry angels also participated.