The Robot Report Podcast · 6 Jan 2021 · 50m
Matanya Horowitz, founder and CEO of AMP Robotics, explains how the company uses AI and custom suction-based grippers to automate recycling sorting, detailing the evolution of their vision system from garage prototypes to industrial scale. He discusses the company's $55 million Series B funding, the impact of COVID-19 on labor shortages, and new data-centric revenue streams for commodity quality assurance. Josh Fox, an applications engineer at Festo USA, describes how his team won the MassRobotics-GM gripper challenge by developing a universal end effector combining Mag Switch magnetic grippers with Festo actuators to handle various sheet metal parts.
Where this conversation describes AI being used to build or deliver a product. Part of the AI in practice tracker.
Engineering & simulationMicrosoft Kinect
The founder used a Microsoft Kinect depth sensor to capture image frames and point clouds of construction materials, creating the initial dataset for training the vision system.
“the 1st version of the system. I was really excited about the Microsoft Connect and its depth sensing. So, uh, the data set was, uh, Yeah, uh, uh, image frames and then point clouds, uh, and, um, you had made my own data set.”
Matanya Horowitz · Hear it at 4:50
Engineering & simulation
The team manually labeled a dataset of scanned materials in a controlled lab setting to train the neural networks, focusing on niches where the vision system showed natural aptitude.
“we did, um, just collect a lot of material, scan it in, um, and then do uh, labeling ourselves. And um, what we kind of did is, uh, so fortunately in the recycling industry, there's a number of different, uh, uh, niches that you can go after. And so we kind of found the ones where the vision system naturally had some aptitude, and then kind of focused on those to show the technology could work.”
Matanya Horowitz · Hear it at 4:50
Engineering & simulation98% purity identification
The company deployed vision systems without robots to analyze commodity purity, allowing recycling facilities to price materials based on exact quality metrics rather than guessing.
“So right now, um, you know, recycling facility, it won't know if its material is is separated out as 95% pure or 98% pure. And there can be real value discrepancies there. So, you know, you put our vision system there and we count everything and we say, okay, you know, 98 out of 100 of those bottles were bottles that you wanted and 2 things were bad…”
Matanya Horowitz · Hear it at 15:58
Engineering & simulationABB Delta robots
The team selected off-the-shelf ABB Delta robots for their pick-and-place hardware, relying on custom vision and gripper design rather than inventing new robot kinematics.
“I believe, uh, Amp uses off the shelf with the Delta robots from ABB and Amron, you must use off the shelf cameras, I would believe, but the the suction gripper is a custom gripper built in house, um, you, you kind of touched upon it.”
Steve Crow · Hear it at 11:42
Engineering & simulationFesto rotary and linear actuators5-6 different parts handled
The Festo team integrated a MagSwitch magnetic gripper with Festo rotary and linear actuators to create a universal end effector capable of adjusting locating pin positions on the fly for different sheet metal parts.
“so, uh, our end effector, uh, the map, uh, mag switch and Festo team came up with, uh, consists of probably 3 main parts … There is a mag switch, uh, E30 magnetic gripper that does all the holding of the workpieces. So obviously we're working with various material. Uh, and that is paired up with a rotary actuator from Festo … and then there's an additional linear actuator that, uh, controls a 2nd 2 way locating pin…”
Josh Fox · Hear it at 38:03
companies scaling up and early-stage companies
Other shows on the same ground, closest first.
Matanya Horowitz, founder and CEO of AMP Robotics, explains how his company uses AI and robotics to automate waste sorting, a sector where only about 35% of US waste is currently recycled. Horowitz details the transition from using mechanical claws to vacuum-based systems to handle sticky contaminants like dried milk, and discusses the economic challenges of making recycling profitable by lowering sorting costs below the value of the recovered commodities. He also describes AMP's strategy of selling robots to existing facilities while building its own green-field recycling plants to demonstrate the technology's potential and drive industry reshoring.
Apurba Pradhan, head of product and marketing at Everest Labs, explains how his company uses AI and robotics to automate sorting at Material Recovery Facilities (MRFs). He details the inefficiencies of manual labor and semi-precise optical sorters, describing how their system replaces human pickers to improve commodity purity. Pradhan highlights the company's service model, which includes 24/7 monitoring and performance guarantees to ensure ROI and address MRFs' lack of technical staff. The conversation covers the impact of the Chinese National Sword policy on domestic recycling demand and the economic viability of automation in the current market.
Areeb Malik, co-founder of Glacier, explains how his company uses AI and robotics to automate sorting at Materials Recovery Facilities (MRFs), addressing the industry's labor shortages and contamination issues. He details the technical challenges of computer vision for trash classification and robotic end-effectors, noting that Glacier's custom-designed robots cost 60% less than competitors and offer a payback period of under two years. Malik also discusses the impact of China's 'National Sword' policy on global recycling markets and the growing demand for recycled content driven by legislation and consumer pressure.
Kim, CEO of Rapid Robotics, explains how the company pivoted from custom manufacturing automation to end-of-line boxing and palletizing using generative AI to solve labor shortages and handle high SKU variability. She details the development of Rapid ID, which allows robots to identify and handle diverse objects without extensive programming, contrasting this with the limitations of traditional robotics in brownfield environments. Kim also shares her background in the toy industry at Hasbro and how that experience informs her approach to scaling the company and supporting US manufacturing reshoring.
Piotr Owczarek, President of Fideltronik, discusses the company's seven-year journey to develop AI-driven robotics for electronic manufacturing services (EMS). He explains how a team of 170 engineers in Krakow is creating flexible automation that mimics human operators to address labor shortages and cost pressures. Owczarek details their 'academy for robots' concept, where collective knowledge is stored in the cloud to allow robots to learn from each other without daily programming. The episode highlights that major EMS companies are currently engaged in pilot programs to validate this technology before full-scale adoption.
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.