Latest news with #JeffMahler
Forbes
9 hours ago
- Automotive
- Forbes
Why Reliability Is The Hardest Problem In Physical AI
Dr. Jeff Mahler: Co-Founder, Chief Technology Officer, Ambi Robotics; PhD in AI and Robotics from UC Berkeley. getty Imagine your morning commute. You exit the highway and tap the brakes, but nothing happens. The car won't slow down. You frantically search for a safe place to coast, heart pounding, hoping to avoid a crash. Even after the brakes are repaired, would you trust that car again? Trust, once broken, is hard to regain. When it comes to physical products like cars, appliances or robots, reliability is everything. It's how we come to count on them for our jobs, well-being or lives. As with vehicles, reliability is critical to the success of AI-driven robots, from the supply chain to factories to our homes. While the stakes may not always be life-or-death, dependability still shapes how we trust robots, from delivering packages before the holidays to cleaning the house just in time for a dinner party. Yet despite the massive potential of AI in the physical world, reliability remains a grand challenge for the field. Three key factors make this particularly hard and point to where solutions might emerge. 1. Not all failures are equal. Digital AI products like ChatGPT make frequent mistakes, yet hundreds of millions of active users use them. The key difference is that these mistakes are usually of low consequence. Coding assistants might suggest a software API that doesn't exist, but this error will likely be caught early in testing. Such errors are annoying but permissible. In contrast, if a robot AI makes a mistake, it can cause irreversible damage. The consequences range from breaking a beloved item at home to causing serious injuries. In principle, physical AI could learn to avoid critical failures with sufficient training data. In practice, however, these failures can be extremely rare and may need to occur many times before AI learns to avoid them. Today, we still don't know what it takes in terms of data, algorithms or computation to achieve high dependability with end-to-end robot foundation models. We have yet to see 99.9% reliability on a single task, let alone many. Nonetheless, we can estimate that the data scale needed for reliable physical AI is immense because AI scaling laws show a diminishing performance with increased training data. The scale is likely orders of magnitude higher than for digital AI, which is already trained on internet-scale data. The robot data gap is vast, and fundamentally new approaches may be needed to achieve industrial-grade reliability and avoid critical failures. 2. Failures can be hard to diagnose. Another big difference between digital and physical AI is the ability to see how a failure occurred. When a chatbot makes a mistake, the correct answer can be provided directly. For robots, however, it can be difficult to observe the root causes of issues in the first place. Limitations of hardware are one problem. A robot without body-wide tactile sensing may be unable to detect a slippery surface before dropping an item or unable to stop when backing into something behind it. The same can happen in the case of occlusions and missing data. If a robot can't sense the source of the error, it must compensate for these limitations—and all of this requires more data. Long-time delays present another challenge. Picture a robot that sorts a package to the wrong location, sending it to the wrong van for delivery. The driver realizes the mistake when they see one item left behind at the end of the day. Now, the entire package history may need to be searched to find the source of the mistake. This might be possible in a warehouse, but in the home, the cause of failure may not be identified until the mistake happens many times. To mitigate these issues, monitoring systems are hugely important. Sensors that can record the robot's actions, associate them with events and find anomalies can make it easier to determine the root cause of failure and make updates to the hardware, software or AI on the robot. Observability is critical. The better that machines get at seeing the root cause of failure, the more reliable they will become. 3. There's no fallback plan. For digital AI, the internet isn't just training data; it's also a knowledge base. When a chatbot realizes it doesn't know the answer to something, it can search through other data sources and summarize them. Entire products like Perplexity are based on this idea. For physical AI, there's not always a ground truth to reference when planning actions in real-world scenarios like folding laundry. If a robot can't find the sheet corners, it's not likely to have success by falling back to classical computer vision. This is why many practical AI robots use human intervention, either remote or in-person. For example, when a Waymo autonomous vehicle encounters an unfamiliar situation on the road, it can ask a human operator for additional information to understand its environment. However, it's not as clear how to intervene in every application. When possible, a powerful solution is to use a hybrid AI robot planning system. The AI can be tightly scoped to specific decisions such as where to grasp an item, and traditional methods can be used to plan a path to reach that point. As noted above, this is limited and won't work in cases where there is no traditional method to solve the problem. Intervention and fallback systems are key to ensuring reliability with commercial robots today and in the foreseeable future. Conclusion Despite rapid advances in digital GenAI, there's no obvious path to highly reliable physical AI. It isn't just a technical hurdle; it's the foundation for trust in intelligent machines. Solving it will require new approaches to data gathering, architectures for monitoring/interventions and systems thinking. As capabilities grow, however, so does momentum. The path is difficult, but the destination is worth it. Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
Mint
4 days ago
- Business
- Mint
The Holy Grail of automation: Now a robot can unload a truck
The robots are coming for the last human warehouse jobs. Loading and unloading a truck is backbreaking, mind-numbing work that retailers and parcel carriers have tried to solve for years. Workers may not stay long in these jobs. Summers and winters are particularly grueling for anyone stuck in a metal trailer, slinging heavy boxes. Injuries are common. Automating this process has long been the holy grail of warehouse logistics. When loaded, packages must be fitted together to fill the available space and be sorted by weight—with the heaviest items on the bottom—so they don't topple or break. Unloading them is challenging, too, because the unloader must move in and out of a trailer, ferrying packages of different sizes and weights. On a typical warehouse floor today, every task might be heavily automated—except for workers loading and unloading the trucks. People who have worked these jobs say they have to stand for extended periods, hefting boxes as heavy as 70 pounds. New advances in robotics are changing that. Improved sensors and algorithms, advancements in AI and faster image-processing technology are making these robots proficient players in tasks that are like a game of 3-D Tetris. Engineers at Ambi Robotics designed a videogame to train its robotic stacking system, AmbiStack. It simulated challenging situations, including heavy parcels and boxes with strange dimensions, said Jeff Mahler, Ambi Robotics co-founder and chief technology officer. Another company, Boston Dynamics, has designed a robot called Stretch, named for its flexible arm that can reach the top corners of a trailer. With a vacuum gripper covered in suction cups, it can lift boxes weighing up to 50 pounds. DHL now has a total of seven Stretch robots in supply-chain facilities in three states and has trained nearly 100 associates to operate them. In Columbus, Ohio, one Stretch robot that DHL staff named 'Johnny 5" unloads around 580 cases an hour, almost twice the rate of a human unloader. DHL in May signed an agreement with Boston Dynamics for 1,000 more robots. United Parcel Service is also increasing automation at its facilities, including for loading and unloading trailers—a move that will help the company cut costs, UPS executives said in April. FedEx has been testing and refining the truck-loading process in one of its facilities with robotics company Dexterity since 2023. Walmart also has introduced robots that can unload a truck. DHL wanted a robot that had the flexibility to handle different products, that could move in and out of a trailer and that didn't require a large capital investment, said Sally Miller, global chief information officer of DHL Supply Chain. 'These are hard jobs to fill, especially unloading a trailer in the warmer months," she said. Stretch the Robot still has some difficulty picking up thin packages, and the robot can't unload bags yet. Boston Dynamics declined to say how much each robot costs but estimates that there is a two-year return on investment on the robots. Megan Diveley, a warehouse worker at a logistics company in Virginia who has been loading trucks for around three years, said she got nasty bruises on her legs when she first started. 'It got better after I got stronger, but I am always peppered with bruises," said the 44-year-old. Diveley said she fears losing her job and feels that worry even without the specter of robots. Lower volume, facility consolidation and cost-cutting at logistics companies are all factors that can result in layoffs. Her advice for the humans still doing the job: stay hydrated and wear steel-toed shoes.
Business Wire
15-05-2025
- Business
- Business Wire
Ambi Robotics Sells Out AmbiStack for 2025 as Fortune 500 Customer Demand Accelerates
BERKELEY, Calif.--(BUSINESS WIRE)-- Ambi Robotics, the leader in AI-powered robotic sorting and stacking solutions for warehouse operations, today announced that its latest product, AmbiStack, is sold out for 2025 following strong market demand from FORTUNE 500 shipping and logistics customers. Since being introduced to the market in January, AmbiStack has received significant interest from leading logistics companies who are ready to embrace next-generation AI technologies that drive increased throughput and multiply productivity across operations. Initial customer deployments of the robotic systems begin mid-year, with 2025 inventory fully reserved. Ambi Robotics is ramping up its production capacity to support significant growth and strategic deployments, and its manufacturing capabilities will be scaled to fulfill demand from customers waitlisted into 2026. 'Developed directly in response to customer feedback, AmbiStack fills a real gap in the market,' said Jim Liefer, CEO of Ambi Robotics. 'Retail and logistics giants needed a smarter, scalable way to handle stacking inbound and outbound packages with speed and accuracy as volumes keep rising, and they needed it a long time ago. Selling out our first full year of product just months after launch shows how urgent the demand truly is.' AmbiStack is an AI-powered robotic stacking solution designed to optimize material handling operations. The solution's multipick capability allows it to pick and place multiple items simultaneously, exceeding manual palletizing rates. Its modular design enables sorting and stacking to multiple pallets with a single machine, reducing the need for constant pallet removal and enabling round-the-clock operations. Pre-trained in simulation, AmbiStack is ready to deploy from day one, adapting seamlessly to different facility layouts. It uses a breakthrough in simulation to reality (Sim2Real) reinforcement learning to stack random boxes with high density. AmbiStack continuously improves its performance post-deployment by leveraging data collected in real-world operations. This data-driven feedback loop is central to the solution's long-term value. To reinforce its commitment to customer data security, Ambi Robotics received its first Service Organization Control (SOC) 2 Type II audit report, which had no exceptions - an added validation of the standards the company continues to maintain. 'As our robotic solutions continue to scale within warehouses, so does the amount of robot data for training foundation models,' said Jeff Mahler, CTO of Ambi Robotics. 'Achieving SOC 2 Type II compliance is a critical step in ensuring secure data management and reinforcing trust with enterprise customers, especially as cybersecurity threats continue to evolve.' Ambi Robotics follows strict protocols for data access, handling and retention to safeguard customer information. The company conducts regular security audits and testing, including third-party penetration tests, to uphold the highest standards of protection. All customer data is encrypted at rest and in transit, and the AmbiOS operating system is hosted across multiple geographically distributed data centers with built-in redundancies. Ambi Robotics supports customer Single Sign-On (SSO) for secure login and uses data for remote support and continuous system improvement, including AI retraining. Ambi Robotics is well-positioned to accelerate the adoption of intelligent automation across the supply chain. As the company expands deployments and scales production, it remains focused on delivering reliable, secure and high-performance AI-powered robotic systems that meet the evolving needs of the world's largest retailers and logistics providers. About Ambi Robotics: Ambi Robotics is an artificial intelligence (AI) and robotics company developing advanced solutions that scale ecommerce operations to meet demand while empowering humans to handle more. The company's industry-leading AI operating system, AmbiOS, leverages proprietary simulation-to-reality (Sim2Real) technology and the latest AI foundation models to provide reliable high-speed robotic systems. At Ambi Robotics, the world's top roboticists, AI researchers, and leading business professionals work together to build cost-effective high-volume supply chain technologies. Founded in 2018, the company is located in Berkeley, Calif. For more information, please visit
