Collaborative Robots Continue to Improve as Amazon Demonstrates the Automation of New Tasks

The Collaborative Robot (cobot) market remains in a turbulent state. While startups bring to market innovation for domestic adoption (Standard Bots and Realtime Robotics in the United States; YardStick Robotics in Germany; COBOTS Inc. in Canada), established vendors grapple with low-cost Chinese innovation. From another dimension, industrial robot Original Equipment Manufacturers (OEMs) (FANUC, ABB, YASKAWA) continue not to regard cobots as serious revenue generators. All concerned are attempting to woo largely apathetic Systems Integrators (SIs) into carrying their cobot products.

Established vendors and innovators continue to iterate on collaborative robots, increasing capabilities and usability. For example, Universal Robots (UR) launched its new low-level torque interface in May, bringing a new dimension to the collaborative capabilities of its robots. UR’s extensive partner network will have access to the new interface, empowering Research and Development (R&D), unlocking new cobot use cases, and inspiring budding startups to enter new verticals. Amazon Robotics launched Vulcan, a UR collaborative robot augmented with touch sensors, earlier in the month. Vulcan’s augmentation will extend the picking and stowing capabilities of robots across Amazon’s warehousing and distribution facilities. Touch and torque control could considerably extend the use cases for collaborative robots—benefits that will filter down to UR’s extensive partner network—further closing the gap between human and machine capabilities.

Amazon’s well publicized adoption of a cobot variant signals to the larger material handling and logistics markets that the technology is viable and here to stay.

One of the greatest shortcomings of robotics is the inability to replicate the dexterous and complex gripping capabilities of human beings. This creates a bottleneck when it comes to picking: many materials, such as fabrics and clothing, lack the rigidity and appropriate surface area to apply vacuum systems and end effector grippers. A human’s tactile sensing allows for sufficient purchase; greater robot force sensor dimensionality will begin to replicate this. Other obstacles to the viability of cobots for 1:1 human equivalence include variability—the current generation of cobots are programmed in the same rigid manner as industrial robots, they are unable to endure external causes of deviation to their paths. A further barrier is the inability of articulated robots to perform “just-in-time” trajectory adjustments for rapidly picking from crowded bins and heterogenous goods. One solution comes from software vendor Vsim offering the market a solution for fine-tuning large machine learning models in real time to account for misalignment between training data and real-world tasks.

Further refinements look to improve either machine vision or control or both. Summer Time Robotics uses a proprietary Artificial Intelligence (AI) software stack to perform rapid classification and path alterations to manage demanding, non-uniform workloads. A similar solution is offered by Micropsi, which offers an end-of-arm tool/controller combination for generating the final adjustments to an articulated robot’s path, enabling complex assembly and inspection tasks. Another startup, Versatile Robotics, provides controller augmentation for safe, stable, and robust deployments, enabling cobots to complete tasks even if the end effector is displaced by a human worker.

New robotic platforms continue to face significant barriers to adoption across both emerging and established market verticals. A key challenge is the operational speed of cobots: the International Organization for Standardization (ISO) limits Tool Center Point (TCP) speed to 250 Millimeters per Second (mm/s). To ameliorate this speed limit, manufacturers often place the cobot in a cage—negating any collaborative advantages. The situation is compounded by the cautious stance of incumbent OEMs, which are often reluctant to disrupt stable product lines in the absence of strong efficiency-boosting incentives. A further bottleneck is the SI layer. Robotics OEMs must not only convince SIs to adopt and deploy complex, often university-spinout technologies, but also mitigate concerns about long-term support burdens and potential risks to reputation. This friction impedes the commercialization of advanced robotics technologies and hampers the rollout of industry-wide innovation.

Despite these challenges, optimism around cobots is well-founded. With buy-in from major adopters like Amazon and enabled by rapid improvements in dexterity, safety, and AI-driven perception, cobots are increasingly viewed as practical, low-risk automation solutions. Their ease of deployment, ability to operate safely alongside humans, and growing vendor ecosystem make them particularly suited for tasks such as pick-and-place, light assembly, and quality inspection. For operations requiring high flexibility—especially in dual-armed assembly (Acumino) or semi-structured environments (Realtime Robotics)—cobots offer a market-ready, reliable alternative to more experimental form factors such as humanoids. As cost declines and capabilities continue to rise, cobots are poised to play a central role in the next wave of industrial and commercial automation.

Another development aiding the uptake of collaborative robots is the buy-in from camera and vision vendors. SOLOMON 3D, Nikon, Mech-Mind, and Robovision all provide tailored, easily configured solutions for inspection and pick and place tasks. These companies target alternative, under-automated verticals (such as agriculture) where the heterogeneity of goods and required dexterity create suitability for cobots. While the use of cobots in manufacturing will continue to grow, their advantages in retail, healthcare, recycling, and logistics will be increasingly recognized and suitable SIs will arise to support these deployments. While inertia surrounds industrial manufacturing—a protected, high throughput industry—automation within new domains will attract innovators with ambitions to capitalize on greenfield opportunities. The symbiosis between cobots and AI has only just begun to materialize.