Robot watching was exciting at CES 2026 – humanoids, pet robots, factory robots, cooking robots, etc. Good to see, but still far away. On the other hand, non-humanoid robots are here and movin’ to propel forward blue-collar autonomy in industrial, security and construction applications. Hyundai Motors unveiled their Atlas robot at CES 2026 (built by Boston Dynamics, which it acquired 5 years ago from Softbank). These are not general purpose robots, and are designed to perform specific tasks like parts sequencing (in 2028), with applications expanding gradually as safety and quality benefits are validated. By 2030, Hyundai projects that these robots will move into component assembly. Beyond this, Atlas is expected to take on tasks involving heavy loads, repetitive motions and complex operations across production sites. But these are projections, not reality today. And Hyundai’s trade unions are not thrilled with this development, stating “its strong opposition to the carmaker’s plan for deploying humanoid robots across its major assembly lines here and abroad”.
As discussed in my previous article, on construction autonomy, a key focus of the Media Panel on The Future of Mobility Innovation at CES 2026 was vehicle autonomy and the emergence of Software Defined Vehicles (SDVs). Another point discussed was that although autonomous vehicles grab most of the public attention in physical AI and AoT™ (Autonomy of Things), the sector delivering immediate business value is blue-collar autonomy – construction, agriculture, mining, airport & warehouse logistics, garbage hauling and critical infrastructure monitoring. L3 and L4 vehicle autonomy are old news and unprofitable. On the other hand, Physical AI and AoT™ advances are adding tremendous value now in blue collar autonomy, for reasons discussed in detail in the construction autonomy article. There is a stark difference between companies working on autonomous vehicles and humanoid general purpose robots vs blue-collar autonomy companies. The former keep promising future roadmaps. The latter are showing impressive operational results now.
This article discusses successful deployment of purpose-built robots for autonomy in factories, security and construction monitoring.
Tata Consulting Services (TCS) – Perpetually Adaptive Enterprise
TCS is a leading IT consulting company, headquartered in Mumbai, India and part of one of the most prestigious industrial groups in India, the Tata Group (which makes semiconductor chips, assembles Apple iPhones, steel, Jaguar-Land Rover cars, other cars, textiles, etc.). It services clients in 46 countries and has revenues of ~$30B, 10% (~$2.5B) of which comes from its manufacturing services consulting arm, which helps clients build a resilient, sustainable, and human-centric manufacturing enterprise. It is headquartered in Edison, New Jersey and headed by Anupam Singhal. Although TCS has served clients in the manufacturing sector for over 40 years, the thrust in physical AI started about a year ago with the realization that their clients needed robots and physical AI to gather and consolidate vast amounts of data in manufacturing and construction environments, and the ability to process this data at the edge to make decisions or issue alerts. According to Mr. Singhal, TCS’s operating model in physical AI is NOT complete autonomy, but based on a HITL (Human in the Loop) architecture. Over the past year, TCS Manufacturing has secured ~15 client projects in physical AI implementations.
At CES 2026, TCS featured a quadruped robot for facility and construction quality monitoring (Figure 1):
The Poochi robot (made by Unitree), is outfitted with a TCS customized sensor and compute stack. Sensors include an IMU (Inertial Measurement Unit), visual (cameras and 360° Field of View LiDAR) and thermal sensors, and TCS developed perception, path planning and decision-making AI enabled software, running on an NVIDIA Jetson processor. It is deployed in multiple applications for sensing, mapping, analysis, failure/defect detection, safety and code assurance. Application examples are:
- Autonomous & Continuous Patrolling in Large Industrial Facilities like factory and campus environments, including areas that are difficult to monitor continuously, and identify anomalies early.
- Quality Inspection in Manufacturing Operations in construction and manufacturing facilities, capturing repeatable visual data across production and utility areas. This improves inspection consistency and enables early identification of deviations without interrupting ongoing operations.
- Safety in Hazardous or Hard-to-Reach Areas like constrained or hazardous zones to remotely assess conditions using built-in safety mechanisms.
- PPE (Personal Protective Equipment)) Compliance Monitoring for Worker Safety by identifying whether workers entering restricted or hazardous areas are wearing required protective gear.
Figure 2 shows examples of the above applications.
A key initiative mentioned during the CES 2026 interview with TCS was use of the Poochi quadruped robot for monitoring construction quality in an upcoming semiconductor plant. According to a recent news report, TCS is teaming up with a sister company, Tata Electronics, in a bid to to accelerate India’s semiconductor production capacity, with upcoming plants in the states of Assam and Gujrat. The Assam plant is expected to be operational this year. Nirmala Seetharaman, India’s Finance and Corporate Affairs Minister recently visited this plant. This is likely to be the initiative that TCS discussed at CES 2026.
According to Mr. Singhal, “Physical AI is where intelligence moves to the edge—into the real world of operations. At TCS, we are enabling manufacturers to extend visibility and decision-making into environments that are difficult, risky, or inefficient for humans to access. Physical AI operates alongside the workforce, extending human awareness and decision-making. The real impact is creating future-ready industrial environments that are safer, more resilient, and continuously aware—at scale”.
Oshkosh – Working for the World
Oshkosh works on autonomy and electrification in multiple arenas. A previous article described the company’s initiatives in construction autonomy. Oshkosh is surgical about autonomy and where it is inserted – harsh environment, blue-collar operations that involve safety and health risks, and improve operational efficiency. The key is transferability of core autonomy building blocks across these different applications. As Jay Iyengar, the company CTO puts it, “We want to develop scalable technology that’s applicable across all our product families, because it’s always easy to do a one-off. It’s a lot harder to do something at scale”. Examples of this in airport outdoor logistics applications are shown in Figure 3:
Apart from the perimeter monitoring and wheel-chocking robots shown above, Oshkosh also displayed at CES 2026, HARR-E, (Hailable Autonomous Refuse Robot, Electric), it’s on-demand, electrically powered garbage collection robot for gated and planned communities (Figure 4):
HARR-E enables residents to request a pickup using a smartphone app or a virtual at-home assistant. Then, HARR-E makes its way to the designated pickup point. It features a two-piece design that makes it easy to lift and transfer waste into a central dumpster or collection point, measuring the volume and weight of waste at each pickup. This data enables it to notify notify waste companies when a dumpster or central container is approaching capacity.
Oshkosh is also developing AI-based tools that identify recyclables from contaminated materials at the point of collection. Integration of this capability in the HARR-E platform will enable segregation of these materials to enable higher recycling efficacy.
Hidonix – Innovating Tomorrow Today
Think about a world where GPS does not work. Quantum sensors are one way to address this gap – due to denied and spoofed environments, coverage gaps, malfunctions, etc. These are great for aerial, maritime, underwater and some terrestrial applications, and are maturing in terms of SWaP (Size, Weight and Power), cost and platform noise mitigation. They typically use geomagnetic field maps (and the use of quantum phenomena) to sense relative and absolute location in the event of a GPS loss, over large geographical areas. Another solution is needed for buildings, arena and campus facilities. This is where Hidonix steps in.
With roots near Sicily (the Italian food capital), the company was founded in 2021 to commercialize indoor mapping for museums (which are GPS-denied) by Achille De Pasquale (the current CTO). It relocated to Santa Monica, California in 2023 to focus on broader applications like security in defense and commercial verticals, using geomagnetic and robotic mapping technologies. A key aspect of their solution is that it enables 3D location mapping (GPS is strictly 2D, think of how to locate someone in a NYC skyscraper – are they on the first or 100th floor?).
Hidonix started off with seed venture funding and government/commercial contracts. It employs ~50 people between California and Italy, and is growing. Defense is a big focus, specifically to enhance situational awareness, decision-making, and safeguard human life in complex environments. The company is understandably secretive about its defense-focused solutions, but emphasizes that the dual-use nature of their technology and products enables commercial markets.
The Hidonix ION (Indoor Outdoor Navigation) solution is an end-to-end platform for high-accuracy indoor positioning, mapping, navigation, and real-time location tracking. The system captures a structure’s location-specific geomagnetic signature to generate accurate 2D and 3D maps through ION’s CMS (Content Management System). Once a space is mapped using a purpose-built robot (Hido II), ION enables navigation services for end users, while businesses can leverage the platform for real-time location tracking, either through smartphones running a white-label ION app or via dedicated tracking devices, including tags and wearables. ION is a dual use system for defense and commercial applications like hospitals, schools, shopping centers, mining sites, and large office buildings, for navigation, asset and personnel tracking, and spatial analytics. (Figure 5):
The mapping is done with the Hido II, a purpose-built, battery operated robot (developed and built by Hidonix). It is an autonomous rover that navigates complex environments to collect geomagnetic data efficiently (and constructed to ensure no interference from the robot platform (see Figure 6):
Hido II integrates a precise sensing stack, controlled mechanical articulation, and a mobility system that maintains continuous movement across complex terrain. It increases acquisition speed by over 50% compared to manual workflows, enabling larger areas to be mapped in significantly less time while preserving consistency in the collected data.
The perception stack consists of a front-mounted 2K stereo camera and a top-mounted Ouster LiDAR that delivers continuous 360-degree point-cloud coverage. Together, these systems enable human and object detection, obstacle characterization, and real-time generation of spatial heatmaps.
The spatial geomagnetic sensing is achieved with a folding, extendable robotic arm that carries the Hidonix Acquisition Module, which includes the HEC (Hidonix Embedded Computer from Nvidia) that runs proprietary firmware designed for precise geomagnetic signal acquisition. The arm’s kinematics support controlled, repeatable placement of the module, ensuring consistent contact geometry and low-variance signal collection across changing environments. The geomagnetic signatures are sensed via magnetometers that utilize the Hall Effect (that detects voltage changes when current carrying semiconductors are exposed to a magnetic field). This is correlated with IMU measurements to provide an exact spatial geometry of the mapped space.
For people and asset tracking, and safety assurance, Hidonix has developed smart wearables (bracelets, badges) to provide continuous positional awareness, powered by the HEC (Hidonix Embedded Computer).
“The value of the ION platform lies in how it comes together – from mapping to navigation to real-time tracking – to give organizations clear, reliable insight into how people and assets move through their spaces,” said Achille De Pasquale, Founder & CTO of Hidonix. “The advancements we’re showcasing this year illustrate how accurate spatial intelligence can improve safety, efficiency, and the overall experience inside complex environments.
Panasonic – The Future We Make
One key aspect of Panasonic’s CES 2026 was its thrust into AI infrastructure across multiple applications like energy storage for data centers, electronic materials for data servers, cooling pumps, compressors, chip packaging equipment and green transformation solutions (solar cells, recycling). An example of the green solutions thrust is the “Circular Economy”, in which home appliances like microwaves and washing machines are designed ground up for robotic disassembly, maintenance, repair and assembly. Figure 7 shows this for a washing machine.
The idea behind the circular economy is to make it convenient for customers to reuse household equipment with frictionless and low cost repair solutions. In some ways, this is antithetical to most companies – they would rather customers throw out the old and ring in the new – but Panasonic seems committed to this initiative, and is investing significantly. Centered around a “Disassembly CPS (Cyber-Physical System)” that visualizes product disassembly ease, the system simulates and optimizes disassembly actions and required time using 3D CAD. The resulting data informs new product design and automated disassembly robots, contributing to efficient part reuse and remanufacturing. Furthermore, precise disassembly enables high-purity separation of materials such as plastics and metals, allowing the production of high-quality recycled materials and advancing the recycling process. By incorporating design principles that enhance maintainability, the Group aims to achieve both longer product lifespans and resource recovery. Panasonic is targeting next-generation recycling infrastructure for a circular society by fiscal year 2028.
The old buzzword for product design was DFM (Designed for Manufacturability) and DFR (Designed for Reliability). With Panasonic’s push, these themes must now encompass robotic operations (rather than manual ones), which dramatically impacts design, product architecture and manufacturing process flows (new suggested acronym-DFRRR – Design for Repair, Reusability And Recycling).
Doosan Robotics – United In Innovation
Doosan Robotics, based in South Korea, was founded in 2015 as a division within the Doosan Group, which is ~130 years old, has 114 global entities worldwide, and revenues of $15 billion in 2025. It is involved in infrastructure support, manufacturing heavy equipment (like excavators, loaders, forklifts through Doosan Bobcat), power generation (turbines, generators), water treatment (desalination), and advanced materials (electronics, fuel cells).
The robotics division develops advanced technology and offers cobot (Cooperative Robots designed to work with people) solutions in manufacturing, service industries and more. Cobots seamlessly work in cooperation with humans, are typically lightweight, designed with human safety and interaction in mind, and not mobile. They have force sensors in their joints that detect obstacles and stop motion to prevent accidents, and do not need specially protected or caged areas to work in. Doosan Robotics currently has the the biggest lineup in the worldwide cobot market with an extensive range of product options, customized for process performance and industrial needs for customers. The newly launched Series-E cobots are NSF (National Sanitation Foundation) certified, and specifically created for the food and beverage (F&B) industry. Figure 8 shows examples of Doosan’s cobot line-up.
The E-SERIES has the flexibility to make virtually any food and beverage with industry-leading safety and hygiene standards. Food hygiene safety is assured with features such as sealed gaps between the cobot’s connecting axes. Low weight and slim design enables easy design and seamless integration into compact workspaces. Its 5kg payload and nearly 3ft reach provides ample capacity to handle any food task with ease. Figure 9 shows an example of the E-Series cobot preparing food like fried chicken and french fries (looks yummy!).
The company plans to continue customizing the E-SERIES for other F&B industry use from baking and barbeque, and frozen food packaging.
Robotics is coming of age and delivering value in blue collar operations. General purpose humanoid robots with human-like intelligence are sexy and newsworthy, but still far away and in the realm of science fiction. Special purpose robots for performing specific blue collar tasks in difficult, repetitive and harsh environments are delivering value now to consumers and manufacturers in terms of quality, safety, cost and process efficiencies.