Profile
The Roboticist Who Went to the Warehouse
The fork goes in at 30 feet. That's where the math starts getting uncomfortable - where a 20,000-pound machine operating at the limits of physics meets a warehouse worker who's been on shift for nine hours. Arshan Poursohi spent years inside some of the most prestigious robotics labs in the world before he decided that the most interesting problem wasn't in a lab at all. It was in a Union City, California warehouse. It was a forklift.
Poursohi co-founded Third Wave Automation in 2018 after back-to-back stints at Google Robotics - where he led the ACME team building 3D environmental understanding systems - and Toyota Research Institute, where he directed engineering for advanced robotics platforms. He'd spent two decades building the kind of perception infrastructure that lets machines understand physical space with centimeter accuracy. Then he walked into a warehouse and talked to the people actually running the equipment.
What he heard changed the product. Warehouse operators didn't want a robot that replaced them. They wanted one that handled the 90% of routine tasks so they could focus on the 10% that actually required judgment. That insight - radical in its simplicity - became the philosophical spine of Third Wave's Shared Autonomy Platform.
"Talking to local operators early on to find out what they need better informs our technology so we can help them and solve their actual problems versus the theoretical problems we imagined in the lab."
The TWA Reach forklift that emerged from that thinking doesn't pretend to handle everything. It operates in four modes: fully autonomous, remote assist, remote operation, and traditional manual. When the system detects a situation outside its confidence threshold, it flags a remote operator - who can supervise multiple forklifts simultaneously through Armada, Third Wave's fleet management system. That intervention becomes training data. The system learns. The ratio shifts. It's a machine that gets more capable every time it admits it needs help.
Investors noticed. Third Wave has raised $82 million, with Woven Capital - Toyota's growth-stage fund - leading the October 2024 Series C. The backer list also includes Norwest Venture Partners, Innovation Endeavors, Qualcomm Ventures, and Zebra Technologies. The Qualcomm and Zebra additions are telling: this isn't just warehouse-robotics venture capital. These are enterprise infrastructure bets.
"We know the world is unpredictable, so we wanted to build a system that knows what it doesn't know so that it can ask for help when it needs to."
The Shared Autonomy concept is grounded in a patent Poursohi holds on "Continual Proactive Learning for Autonomous Robot Agents" - the idea that a robot system should actively seek out its own knowledge gaps and use human input to fill them. It's a direct counter to the tech industry's habit of shipping systems that fail silently. Third Wave's forklifts fail loudly, on purpose, and get better for it.
Before the Forklift
Before Poursohi arrived at warehouse problems, he was working on considerably smaller ones. His early career at Sun Microsystems Laboratories put him on Project Sun SPOT - a wireless sensor platform that packed a Java-programmable computer onto a circuit board roughly the size of a credit card. The project was about networked physical devices long before "Internet of Things" became a marketing term. That work built his intuition for distributed systems and sensor fusion: the same foundations underneath an autonomous forklift fleet.
At Google Research and then Google Robotics, Poursohi worked on cloud robotics and object recognition before founding the ACME team, which focused on spatial understanding - giving robots the ability to build and navigate mental maps of physical environments. The technical vocabulary he developed there (LIDAR fusion, 3D semantic mapping, edge-case detection) is now the vocabulary of Third Wave Automation's product specs.
His undergraduate degree from UC Berkeley was in Cognitive Science - a discipline that sits at the intersection of neuroscience, psychology, and computer science, and one whose core question is: how does intelligence actually work? It's not the obvious background for a warehouse robotics CEO. But the question Poursohi is answering - how does a machine know when it knows enough? - is arguably the same question Cognitive Science has been asking for fifty years.
The Infrastructure Swap Nobody Wants
One of Poursohi's most frequently cited observations gets at why warehouse automation is so hard. "You can change warehouse infrastructure," he's said, "but you have to be careful about it. It's like swapping out the engine on a plane in mid-flight." Warehouses aren't greenfield deployments. They have existing racking, loading dock schedules, workforce workflows, and SKU configurations that can't pause for a robotics retrofit. Third Wave built the TWA Reach to work in existing warehouse layouts - no infrastructure modifications required.
"If you automate one piece of operations without thinking about the neighboring sections, you might be shifting a heavy load to a person instead of making a better system overall."
That constraint - design for the world as it is, not as you'd like it to be - runs through every product decision. The forklift uses automotive-grade 3D lidar, the same sensor class that autonomous car programs spent billions developing. Third Wave is repurposing that investment for 10-foot warehouse aisles. The Collision Shield obstacle detection system provides 360-degree visibility. The perception pipeline operates at centimeter accuracy - tight enough to navigate high-density racking without margin for error.
What's Next
Third Wave's Series C signals a shift from development to scale. The company was on the floor at Modex 2024 in Atlanta - the material handling industry's flagship event - demonstrating the TWA Extended Reach platform live. The Armada Fleet Management System, launched in mid-2024, lets a single operator manage multiple autonomous forklifts simultaneously, which changes the unit economics of deployment significantly.
Poursohi's long-game aspiration is a fleet that accumulates intelligence across all deployments - each edge case handled on any customer's floor becomes training signal for every system in the network. The more warehouses Third Wave operates in, the smarter every forklift becomes. It's a compounding advantage that looks modest on day one and formidable at scale.
The warehouses are already running. The forklifts are already learning. The engine swap is underway.
