Benji Barash

MassRobotics is excited to announce the second cohort of the Physical AI Fellowship powered by @awscloud Startup and @nvidia Inception. 🤖 2026 Cohort: @burro_ | @config_inc | Deltia | @HaplyRobotics | Luminous | @roboto_ai | Telexistence | Terra Robotics | WIRobotics Inc. Learn more here: massrobotics.org/physical-ai-fe… @AWSstartups | @NVIDIARobotics

We're thrilled to have been selected for the #PhysicalAIFellowship by @awscloud @nvidia @MassRobotics. Excited to join some amazing robotics companies like Burro, Luminous, Haply and more! Roboto is used by teams building complex systems, including drones, surgical robots, and autonomous vehicles, where reliability and uptime are critical. We’re excited to use these resources to accelerate Roboto’s development and expansion, giving our customers the insights they need to scale 🚀

Last night, @Unusual_VC & @roboto_ai hosted our 2nd dinner on Scaling Reliability in High-Stakes Robotics🤖 From surgical robots to autonomous forklifts, industrial arms to flying taxis, we saw different domains but the same reality: As robots scale from pilots → fleets, the long tail of edge cases + operational complexity become the problem. We dug into: 1️⃣ End-to-end learning in regulated, safety-critical systems 2️⃣ Detecting out-of-distribution events at massive scale 3️⃣ How connectivity tailwinds (e.g., Starlink) are unlocking reliable edge data Grateful to the incredible group of robotics leaders who made it such a rich conversation.

Robotics is absolutely having a moment, but flashy demos don’t automatically turn into fleets that scale. I was on @NYSE Live with @KristenScholer last week, and we dug into this as part of their #CES coverage. One of the toughest challenges in robotics today is going from a few robots to a large fleet. You might have never noticed a problem that shows up once in a thousand missions. But as soon as you have a fleet of 100 robots, it might start happening multiple times a day. The long tail of bugs and edge cases quickly adds up, and finding them is painful and expensive. If the industry can’t figure this out, #robotics will continue to struggle. This is why we’re building @roboto_ai We’re helping teams automate analysis of their robotics data so they can catch issues before their customers do, and scale up their fleets with confidence. Let's make 2026 the year of real fleets!

Today on NYSE Live | It's an all-tech episode. @BabcockWilcox CEO reveals how the firm is built to thrive in the AI data center era. @DivesTech and Cullen Rogers join live after @Wedbush rings the Opening Bell! Plus, hear from @roboto_ai, @_aiOla, & more twitter.com/i/broadcasts/1…

@hans_ojisan haha yes!! I first showed it at the demo stage in Odense - thanks for sharing 🤗

I love this slide since I saw it in #ROSCon 2024!

I showed tools we're building at @roboto_ai to help: 1️⃣ Signal similarity search for pattern detection 2️⃣ Vision-language models for event annotation 3️⃣ AI summarization for faster triage And a sneak peek of an AI agent that debugs robot logs on its own 🤖 @OpenRoboticsOrg

Just wrapped up my talk at #ROSCon in Singapore! 🇸🇬 Robotics teams are drowning in logs - the real challenge is finding the important events hidden inside! #Robotics #ROS #AI #RoboticsData

Our team is in Singapore 🇸🇬 this week for #ROSCon2025 — come say hi at Booth 19/20 for stickers and our famous socks. Our CEO, @BBarash, will present in the main conference tomorrow at 4.30pm: “Detecting Complex Events in ROS Data” 👀👀

Live from #ROSCon in Singapore 🇸🇬! John Hubberts on our team gave an awesome workshop on how to automate #ROS bag analysis with Roboto. The room was packed! #ROSCon2025

Catch us at #RoboBusiness this week at booth 816. Our founder @BBarash will also be diving into the two data problems of #robotics and how new data curation techniques can help solve them. #data #PhysicalAI @NVIDIARobotics

Last night, @Unusual_VC and @roboto_ai hosted a thought leadership dinner on “Scaling Reliability in High-Stakes Robotics.” Leaders @Waymo, @amazon & more joined to tackle challenges like scaling fleets, foundation models, and data flywheels. Excited to keep the convo going! 🤖

We've had an overwhelming response to our Robotics Happy Hour for @Techweek_ today at 4 PM, with almost 400 requests to attend. We're bringing together a curated mix of top startups and investors shaping the future of robotics and AI. Our guest list includes engineers and founders from Anthropic, DeepMind, Apple, Tesla, Waymo, Meta, and NVIDIA, alongside cutting-edge startups like Agility Robotics, Chef Robotics, and Verne Robotics. Leading investors from 500 Global, Samsung Ventures, GS Futures, and Lowercarbon Capital are also joining, plus researchers from Stanford, UC Berkeley, and University of Tokyo. It’s shaping up to be an incredible night of innovation and collaboration in the Bay Area robotics community. Link to RSVP in the comments! Panel starts at 4:30 PM with: - @emilyyu1217 from @BoostVC - @Abhindas1 from @OrangewoodLabs - @BBarash from @roboto_ai I'm still adding folks to the RSVP, so get on the list if you haven't yet! If you're not approved but think you should be, DM me. RSVP HERE: partiful.com/e/6Sir9B3OIvSG…? #SFTechWeek

A few pics from last week! Super fun to catch up with friends and meet a ton of robotics companies. We had an #ANYmal robot at our booth thanks to the amazing folks at @anybotics 🤩

Can AI help debug a drone crash? 🤔✈️ Like all robots, drones generate massive amounts of multimodal log data every mission. These logs are essential for diagnosing issues, but reviewing them is complex and time-consuming....🧵

Huge news for our friends at @allen_ai developing truly open source models, including for robotics! Roboto got started in their incubator program.

@IlirAliu_ hit the nail on the head here, recommended reading for anyone in robotics🧵

better data infrastructure. The companies building this foundational layer are the hidden titans of the industry. They are turning the data chaos into a competitive advantage. The teams that master their data will win. The rest will be left debugging.

This isn't just about better tools. It's about velocity. The teams that master their data can iterate, debug, and deploy updates in hours, not weeks. They can build smarter, safer AI because they can actually learn from their mistakes at scale. The future of robotics isn't just better hardware. It's…

"Show me every time the robot failed to grasp an object in the last 3 months." Hours of searching → seconds of querying. The third breakthrough: