The Robot Report Podcast · 19 Jul 2024 · 1h 26m
Max Cao, CEO and co-founder of Jacobi Robotics, explains how his company uses AI-driven motion planning to simplify robot programming, highlighting their 'Jacobi Paletizer' application and a 95% reduction in commissioning time. The episode also features Jan Hennecke, product manager at Igus, discussing the 'RBTX' low-cost automation platform, which leverages plastic-based components to make robotics accessible to small businesses and educational institutions. Additionally, the hosts discuss industry news, including Agility Robotics' 'Digit' humanoid pricing and funding, as well as the rise of robotics foundation models from companies like Skilled AI and Helm AI.
Where this conversation describes AI being used to build or deliver a product. Part of the AI in practice tracker.
Engineering & simulationJacobi Robotics (Jacobi Planner)95% reduction in commissioning time
The company used AI-based motion planning to calculate optimized, collision-free robot trajectories in real time, replacing manual teach-pendant programming and reducing commissioning time by 95%.
“Our partners have benefited from, for example, a 95% reduction in robot commissioning time, and a 24% saving in total project cost, including hardware.”
Max Cao · Hear it at 31:08
Engineering & simulationJacobi Robotics (Jacobi Planner)1000x improvement in speed
The team developed a deep reinforcement learning framework with accurate robot dynamics models to generate and validate trajectories in nanoseconds, achieving a 1000x speed improvement over existing algorithms.
“We then embed these models into a deep reinforcement learning framework where the agent can generate and validate trajectories and nanoseconds. … this is the, you know, fastest ever measured for collision-free motion planning and as a 3 order magnitude or about a 1000 X improvement compared to existing algorithms.”
Max Cao · Hear it at 39:52
Engineering & simulationSkilled AI (Skilled Brain)1000x more data points
The startup developed a robotics foundation model trained on 1000 times more data points than competing models to provide baseline manipulation and navigation functions for diverse robot embodiments.
“Skilled is trying to differentiate by claiming that, you know, they're training its model on at least a 1000 times more data points. than competing models.”
Mike Oitsman · Hear it at 24:29
early-stage companies and companies scaling up
Other shows on the same ground, closest first.
Misa Ilkhechi, founder of Formic, explains how 'robots as a service' shifts automation from CapEx to OpEx, allowing manufacturers to pay for output rather than owning hardware, a model particularly suited for small-to-medium enterprises. Joe Campbell of Universal Robots highlights the industry shift toward collaborative robots and the enduring necessity of process knowledge, while Etienne Lacroix, CEO of Vention, describes his platform as 'industrial Lego' that democratizes machine design through a browser-based user experience. Finally, Kel Guerin, CIO of Ready Robotics, details their software approach to simplifying robot programming via flowcharts, aiming to empower non-technical factory workers to deploy automation without specialized coding skills.
Josh Gruenstein, co-founder and CEO of Tutor Intelligence, explains how his MIT spinout uses AI and machine learning to automate high-mix, low-volume manufacturing environments, such as co-packers and 3PLs, which were previously considered too variable for robotics. He details how their robots eliminate the need for specialized programming by interpreting build sheets and specifications directly, allowing facilities with no automation engineers to adopt automation. Gruenstein also describes Tutor Intelligence's 'Robot as a Service' model, which converts capital expenditures to operating expenses, reduces risk, and aligns vendor incentives with customer productivity through continuous software updates.
Etienne Lacroix, Founder and CEO of Vention, explains how his company is democratizing industrial automation for small and medium-sized manufacturers by productizing robotics applications and enabling shop-floor practitioners to build their own machines. He details the shift from expensive, custom system integration to modular, plug-and-play components and virtualization, which allows users to create digital twins and de-risk capital expenditure before buying hardware. Lacroix also discusses the recent decline in US robot arm sales, attributing it to interest rates, and outlines how AI-driven tools like generative design and autonomous robotics are lowering the barrier to entry for automation.
Andy Lonsberry, Founder and CEO of Path Robotics, explains how the company identified a market pain point in high-mix, low-volume welding and developed a fully autonomous robotic system using machine learning and computer vision. He details the journey from a 24-month consulting exploration to securing $15 million in Series A funding by bringing an autonomous robot to Silicon Valley. Lonsberry also discusses the 'Robots as a Service' (RAS) business model, which eliminates upfront capital expenditure for manufacturers by tying payment to system performance, and highlights the abundance of engineering talent in the Midwest.
Nick Thompson, CEO of BOW, explains how his company’s SDK abstracts the complexity of robot programming by providing a unified, hardware-agnostic layer that works across simulation and real-world hardware. He details how BOW complements ROS by handling low-level communication and inverse kinematics, allowing developers to write applications that are decoupled from specific robot form factors like quadrupeds or drones. Thompson highlights a key use case with the ORE Catapult’s Cyrus project, where BOW unifies aerial, underwater, and surface vessels for offshore wind inspection, and outlines plans for an App Store marketplace to monetize reusable robotic skills.
The summary, chapters and topics are generated from the episode’s audio by this directory, not written by The Robot Report Podcast. The episode itself belongs to its publisher.