Trailer loading and unloading remain the most difficult processes to automate in the warehouse because every load, trailer, and dock presents different operating conditions, noted Gideon Brothers. The company said its TREY autonomous forklift can both load and unload every pallet on a trailer and is now available in North America.
“Supply chain operators have been missing out on dock automation simply because automated forklift technology had not delivered on the critical requirements every warehouse needs – the ability to safely load and unload trailers,” stated Edin Kočo, Ph.D., co-founder and chief robotics officer of Gideon. “Gideon’s TREY automated forklift has mastered trailer loading and unloading because it possesses advanced physical AI technology to recognize and respond to more variability on the dock than any other solution on the market.”
“TREY is the first critical step in loading bay automation, connecting key modalities — transportation within the four walls of the factory or distribution center on the one side, and loading and unloading trailers on the other,” he told Automated Warehouse.
Founded in 2017, Gideon said its physical AI platform can help supply chains become safer, more predictable, and free from the constraints of manual labor. The company has headquarters and testing facilities in Bolingbrook, Ill., and Osijek, Croatia. It has $75 million in venture funding and has a long-term partnership with Toyota Industries.
“In October 2025, we opened a new manufacturing facility, and we’ve reached capacity,” said Kočo. “We’re establishing partnerships to further accelerate and operationalize growth, allowing us to support customers from operations and automation deployment to post-deployment services.”
TREY built to scale with demand
Gideon said it designed TREY for warehouses that need automation to scale to meet high-volume demand. It claimed that the automated forklift performs at its peak on loading docks that receive 30 or more trailers per day.
“Over 70% of trailers have packaged goods with load carriers are loaded and unloaded by forklift-style vehicles , not loose-loaded goods like those handled by Pickle or Boston Dynamics’ Stretch,” Kočo said. “What allows us to do it is our flexible navigation, and our end effector has four degrees of freedom, almost like a mobile manipulator. It is able to push against a load or the walls of a trailer, mimicking how humans operate today.”
The company added that TREY can improve operational consistency and resiliency on the dock with the ability to:
- Move up to 28 single- or double-stacked pallets or cages per hour, or 224 per eight-hour shift.
- Identify and handle more than 119 load-carrier types of different sizes and weights.
- Work with various trailer styles ranging from dry van, refrigerated, drop deck, intermodal, and more.
- Charge autonomously between cycles, extending operating time across multiple shifts. It can serve both inbound and outbound workflows.
The autonomous mobile robot (AMR) is available in medium- and heavy-duty models to match the demands of manufacturing plants or warehouses. The heavy-duty model can move payloads of industrial parts and components weighing up to 6,000 lb. (2,721.5 kg). It has an operating time of 10 hours on one battery charge.
The medium-duty model can carry a payload of 3,500 lb. (1,587.5 kg) and operate for 12 hours on a single battery charge. TREY has already been deployed in automotive factories, food and beverage distribution centers, and third-party logistics (3PL) providers.
Gideon creates physical AI
Gideon said TREY operates on its proprietary physical AI platform, which uses image-based perception technology and sensor fusion to interpret its surroundings. The company’s vision module, designed in-house, uses a two-camera stereoscopic system to gather depth information and provide context about objects in the path or perimeter of the forklift.
“The unloading side typically doesn’t know what happened on the loading side, since they’re at two different facilities, and the companies are not necessarily connected,” said Kočo. “This means robots have to be able to quickly understand what they perceive, decide, and act in the world.”
“Systems that only use 3D lidar can determine an object’s location and distance away,” he said. “Image-based perception adds meaning and context to recognize the type of load carrier, including its orientation and whether it needs to be repositioned before loading or unloading so that TREY knows what to do next.”
The system uses neural networks, machine learning, and its vision module to recognize and respond to changing conditions, including:
- Dunnage – The forklift can detect bulkhead insulation, air, or dunnage bars or belts in the trailer and notify an associate to remove it. When dunnage is needed during loading, TREY pauses outside the trailer until an associate adds it and confirms that the robot can proceed.
- Orientation patterns – Inbound trailers arrive with pallets in varying arrangements. TREY can recognize different orientations and rotate pallets as needed during unloading. For loading, multiple patterns can be preconfigured in advance for a fast and efficient process.
- Other objects – TREY’s perception system recognizes more than 30 other objects common on the warehouse floor, such as equipment, people, racks, and barriers. Its real-time autonomy software uses the information to plan movements on the floor, improving safety and reducing product damage and loss.
- Dynamic staging – When the dock is congested, the autonomous forklifts can share staging areas with manual forklifts, instead of stopping and waiting an entire trailer cycle to restart. As pallets move and space opens up, automated loading or unloading continues without waiting for the full lane to clear.
“Exposure to real-world operations is critical,” said Kočo. “We use simulation to bring our systems to certain levels of maturity, but we’re truly polishing our models through real-world exposure. Our training process is very simple, and inference runs on the edge.”
“The data pipeline is more complex because we collect every interaction between the robot and a pallet,” he continued. “Data captured with sensors is eventually shared with our servers to continue training our system.”
Sensors provide 360-degree awareness for safety
“According to OSHA, the loading bay is one of the most critical environments. Two out of five serious injuries happen there,” noted Kočo. “TREY can safely handle standard and nonstandard pallets, racks, and containers.”
TREY uses two laser scanners to create a 360-degree safety field around it, while safety zones automatically adjust to changing conditions. An embedded safety controller continuously monitors safety-related inputs and can trigger immediate stops. The forklift is equipped with six emergency-stop buttons plus a remote-control stop button.
The system complies with the industrial vehicle safety standards including ANSI/ITSDF B56.5-2019 and ISO 3691-4:2023.
Trey touts consistent dock automation
No expensive changes to the facility, dock, or trailers are required to implement TREY autonomous forklifts, according to Gideon. Every deployment comes standard with fleet management software and an analytics dashboard that give dock managers tools to validate the automation’s performance.
- Fleet management software coordinates the robots as they serve multiple dock doors in the facility. The software integrates with warehouse management systems (WMS), warehouse execution systems (WES), or stands alone.
- The Fleet Intelligence Dashboard collects and presents operational data from the automated forklifts, including how many trailers have been loaded or unloaded, the speed and performance of the robots, and the number of associate interactions needed to support the work. Warehouse teams can use the information to compare performance between shifts, recognize patterns, or identify training needs.
- As pallets are loaded and unloaded, TREY captures time-stamped images that provide an audit trail for the inventory.
“Variability was essentially a bottleneck for automation, forcing it to be deployed in limited scenarios,” said Kočo. “Customers need robots that can operate across a high number of different commodities. The loading bay is a contact sport — businesses need to be able to manipulate goods while maintaining their integrity, reducing damage, and retaining flexibility.”