America in Motion founder explains how machine learning, vision and speech on a universal PC-based platform can move mobile robots forward for intralogistics
Tommy Hessler, CEO of America in Motion (AIM), has worked with automatic guided vehicles (AGVs) since the mid-1980s. During that time, the AGV industry has repeatedly forecast exponential growth but has not widely embraced leading-edge technologies, says Hessler, who has assisted the National Institutes of Standards and Technology (NIST) with defining AGV standards. However, Charlotte, N.C.-based AIM has always taken a different route.
Since Hessler and Executive VP Theresa Blasius founded AIM in 2007, the company has worked on numerous custom AGV projects using Beckhoff Automation technologies. AIM uses various PC-based control and EtherCAT I/O hardware, along with TwinCAT automation software for supervisory control or navigation in certain cases. The company’s early automatic guided cart (AGC), built more than 10 years ago, ran TwinCAT 2 on a DIN-rail-mounted CX1010 Embedded PC with 500 MHz of processing power. Now, AIM uses more powerful CX5120 Embedded PCs and CP6606 Panel PCs as an operator interface to control its call stations. Hessler is also beginning to implement advanced IoT capabilities using TwinCAT 3, many-core control via CX2042 Embedded PCs with Intel® Xeon® processors and other future-facing technologies to turn AGVs into next-generation mobile robots.
PC Control magazine recently spoke with Hessler about the future of AGVs and the industry.
What trends and technologies are changing the course of the AGV industry?
Tommy Hessler: In the broader industrial world, many manufacturers and OEMs are embracing Industrie 4.0 and IoT concepts. But in the world of automatic guided vehicles, this has not hit home yet — despite how smart and capable Industrial PCs (IPCs) have become. America in Motion is adopting these new ideas and upgrading AGV controllers and software platforms to implement Industrie 4.0 concepts. We are using PC-based control from Beckhoff to increase functionality, for example, by adding advanced speech, vision, cloud connectivity and machine learning capabilities. As a result, AGVs are becoming a PLC on wheels.
How are customers influencing these changes?
Tommy Hessler: Our customers want to collect actionable information from every aspect of their business. Because AGVs drive around the plant all day, they offer a valuable way to collect operations data as mobile IoT devices. However, the AGV control platforms widely available today do little more than navigation, and they rely on outdated communication via CANopen or another legacy fieldbus. The system openness and multi-protocol communication capabilities available through the EtherCAT industrial Ethernet system help us incorporate CAN devices now, but we would like to use only EtherCAT in the future as we standardize on Beckhoff.
The capabilities of TwinCAT and other Beckhoff technologies allow us not only to collect information but also to present it using via HTML5-based TwinCAT HMI software and standard web servers. With the Beckhoff platform, we can do this without writing extensive code, which significantly reduces programming effort. Also, the integration of Microsoft Visual Studio in TwinCAT 3 makes programming easier for us, since we are most experienced with C# and JavaScript. So if there are 10 steps, for example, to reach our goals, we are essentially starting at step seven by implementing Beckhoff technology.
What data types are most useful to distribution center and intralogistics managers?
Tommy Hessler: There are three main categories. The first is performance data, which means the measurement of how many loads are moved per hour per AGV. Beckhoff offers much greater capabilities to collect and graphically present this data.
The second category is safety. These autonomous vehicles move around busy facilities among people, using light curtains, safety bumpers and other safety devices, per ANSI and ISO standards. We want to collect information any time these safety devices are triggered and take a photo, automatically blurring faces to protect privacy, to determine what happened. There are black-box solutions for this that require the AGV to manually plug into a separate computer. We are developing solutions that will tie the functionality into our overall control infrastructure that will run on a CX2042 Embedded PC and transmit data wirelessly.
The third category covers system performance, which means keeping statistics on routes and efficiency changes by day or hour. This allows facilities to optimize AGV routes with incredible results. For example, I performed a study for a dominant beverage producer on one warehouse: Without optimization, it required roughly 180 AGVs, but once optimized with better data collection and analysis, it only needed 32. So with this data, the efficiency increases are substantial. Soon, this optimization will leverage neural networks, such as those made possible by TwinCAT Machine Learning. There again, Beckhoff helps us start at step seven.
Beyond functional safety, do you foresee TwinCAT Vision enhancing AGV systems?
Tommy Hessler: I see software for image processing, such as TwinCAT Vision, as enabling technologies for AGVs to find pallets and racking more efficiently using GigE cameras. Today AGVs use lasers or a guide taped onto the floor to determine their X-Y potions in warehouses, but they can’t identify specific objects or racks. This requires time-consuming measurement, and if anything changes, engineers have to return to measure again. On top of that, if customers want to move the guide paths even slightly, they often don’t have the proper tools to do so. Some AGV vendors play games and do not provide customers the rights to a facility’s definition tables. This isn’t their proprietary information; it’s simply your floorplan.
By implementing 128-beam lasers that update 50 times per second with integrated vision and machine learning, the AGV could basically survey and map the distribution center without human involvement, adjusting if shelving positions or heights change. A pallet could be located anywhere, but once it is identified, the AGV calculates the path and simply picks it up on its own. With open, PC-based control technologies in a single platform, customers can better manage their data and operations.
Are warehouse workers and managers excited or apprehensive about these technologies?
Tommy Hessler: Many times selling AGVs seems like selling a car in the 1890s: The technology is already unfamiliar, and if you started to talk about things way ahead of their time, like CD players, no one would understand anything. Management wants to improve throughput, but doesn’t know why they need an AGV, how it works or if it’s safe, even though it has light curtains all around. We make an effort to guide everyone through the experience and give AGVs more familiar, human attributes. We make the vehicles unique colors and give them names, like Frank and Dottie. Now, using TwinCAT Speech, we are designing a better system, especially for picking operations. Instead of typing ‘I’m done’ or pushing a button after every pick, operators with a headset will simply talk to the AGV by name and have both hands available to pick boxes at the next stop.
Having a mobile robot by your side that listens and verbally responds to you will enhance operations and make U.S. businesses more competitive for workers and management, and that’s what we want. People often feel automation takes away jobs, but we named our company America in Motion because we want to do the exact opposite. Our vision is to make robots friendly so they really collaborate and add value rather than compete against humans. Implemented correctly, we can keep the whole production and desirable jobs here in America. With the TwinCAT software platform and IPC hardware from Beckhoff, we can continue innovating to keep manufacturers and distribution centers moving forward at full speed.
Are you interested in implementing new control and networking technologies in your AGVs or other equipment? Contact your local Beckhoff sales engineer today.
James Figy is the Marketing Content Leader for Beckhoff Automation LLC.
A version of this article previously appeared in Machine Design.