NEURA Robotics connects cognitive arms, mobile robots and humanoid development

NEURA Robotics develops robots that combine physical movement with sensing and AI. Its portfolio spans industrial arms, mobile transport and humanoids, but these products solve different jobs and have different buying routes. A factory should choose a defined operation before choosing a robot shape. This public-source review proposes an industrial evaluation; it does not report hands-on robot testing.

01 / ProductA robotics portfolio with several levels of autonomy

MAiRA is NEURA’s cognitive industrial robot. The company describes a seven-axis arm combining cameras, spatial audio and force-torque sensing, with on-device processing for multimodal inputs. The practical idea is to respond to more than a pre-taught position: perception can inform what the robot handles, while force feedback can inform contact. Those are product claims, not evidence that an arbitrary task will work without integration.

LARA is the collaborative-arm family, positioned around industrial precision, visual programming and prepared applications such as palletizing and welding. It is a more bounded automation proposition than a general humanoid assistant. Buyers should compare the actual application package and work envelope, rather than assume every NEURA arm inherits all MAiRA sensing features.

MAV covers autonomous material transport, including navigation choices and fleet operation. 4NE1 extends the portfolio into humanoids and includes remote-operation capabilities and alternative configurations. The current humanoid page invites reservations and model inquiries. Treat that route as a discussion about a specific deployment, not proof of unrestricted immediate delivery or autonomous competence in every advertised environment.

02 / AudienceIndustrial teams should start with the contact task

A strong NEURA candidate is a manufacturer that can identify why conventional positioning alone is insufficient. Parts may arrive slightly displaced, people may share a work area, or material transport may repeatedly interrupt a productive station. Those are different problems: an arm can manipulate a workpiece, while a mobile platform can deliver it. Buying both creates an additional handoff that needs its own definition.

For a stable machine-tending cell, compare LARA’s application tooling with MAiRA’s sensing proposition before paying for a broader capability. For a research team exploring human-shaped manipulation, 4NE1 may be relevant, but its reservation conditions and supported software need separate confirmation. A household buyer should not use an industrial-arm specification as evidence of a ready consumer service.

The Teradyne blueprint provides context on collaborative robot and mobile-robot platforms with application ecosystems. The Figure blueprint is a more relevant comparison for humanoid development. This division helps avoid comparing an industrial workcell purchase with a research or early-deployment relationship simply because both involve AI-enabled robots.

03 / WorkflowA proposed machine-tending trial should include the handoffs

Consider a proposed MAiRA evaluation that transfers small manufactured parts between an input tray and a machine fixture. First define the part family, tool, fixture and supported variation. Include the parts that consume operator time today, such as pieces that sit differently in a tray. Exclude unknown geometry only when it is explicitly outside the proposed application, and record that exclusion.

Have the application engineer explain which observation drives each action. A camera may establish a candidate position, while force sensing may determine whether contact or seating has occurred. Ask for the actual completion signal used by the installed application. A robot reaching a coordinate is not the same thing as a correctly seated part, and visual recognition alone cannot establish the quality of every mechanical fit.

Build the proposed sequence around load confirmation, machine readiness, manipulation, verification and a defined release to the next stage. Interrupt the sequence at ordinary handoff points: a fixture remains occupied, a part is missing or the machine has not issued a ready signal. The purpose is to learn whether recovery preserves the production state. Repeating a motion after an uncertain placement can compound the original problem.

If material transport is the bottleneck, evaluate MAV separately with the actual aisle geometry, load carrier and destination station. Its documented navigation alternatives include Navitec and Node Navigation. Ask which is included in the proposed configuration and how an occupied destination is handled. A vehicle arriving nearby does not automatically complete unloading or make the next machine available.

Include operator intervention in the timing. Record correctly completed parts, assisted recoveries and time spent replenishing the cell. Review sustained output over a realistic production window, including changeovers. This is a suggested acceptance method, not a claimed NEURA benchmark. It makes the result useful to the production team that will inherit the cell.

Finally, test a deliberate product change within the agreed family. Determine what is parameter editing, what requires a new skill or tooling change, and who can authorize it. NEURA’s descriptions of intuitive interfaces and prepared applications are reasons to investigate changeover effort; they do not establish that every new workpiece can be introduced without engineering.

04 / PricingIndustrial supply uses offers and delivery terms

The public business contact route separates business and private inquiries. The inspected product pages do not provide a universal installed-system tariff. The linked general delivery terms apply to business buyers and make the order confirmation central to the products supplied. Individual agreements can take precedence.

Those terms generally place prices in NEURA’s offer, exclude VAT and describe ex-works supply with ordinary packaging. Shipping, special packaging and related charges need attention in the quotation. The default payment split is 50% within 14 calendar days of invoice receipt and the remaining 50% within 14 calendar days after delivery, unless separate terms apply. These are published default conditions, not a quote for a particular robot.

The meaningful industrial comparison is therefore a specified workcell: arm, sensing, tool, machine interface, installation, training and support. Separately, the 4NE1 Gen 3.5 reservation page lists estimated unit prices of €98,000 for 1–19 units and €60,000 for 20 or more, excluding taxes and shipping. Its €100-per-unit reservation fee is credited to purchase and refundable before signing the final purchase agreement. Reservations are intended for individual orders; commercial quantities use the contact route. The stated expectation of end-2026 availability is not a delivery guarantee or current-stock claim.

05 / DistinctionsSensing and robot form are separate advantages

MAiRA’s distinction is the combination of sensory inputs with manipulation in an industrial arm. The useful question is which inputs improve the selected operation. Speech interaction might help an operator choose a workflow, while force feedback may matter during contact. A buyer should ask for evidence at that level rather than treat multimodality as a single measurable performance benefit.

The broader portfolio offers a possible path from a fixed task to material movement and more flexible manipulation. It can also create complexity if a team tries to automate too many handoffs at once. Starting with one bottleneck helps establish whether additional mobility creates usable capacity. This is an implementation choice, not a claim that one robot form is inherently superior.

LARA’s prepared applications may be valuable when the process is already well understood. In that setting, a repeatable application package, clear recovery behavior and available service can matter more than the breadth of the AI story. The relevant distinction is how much of the production task arrives supported and how much remains an integration project.

06 / QuestionsAvailability and safety claims need configuration-level evidence

The 4NE1 page still refers to first Mini deliveries in spring 2026 while presenting a reservation route. On this October review date, that wording does not establish current stock or a delivered installed base. Obtain a current shipment schedule for the exact model. The page also describes multiple configurations, so specifications from one should not be carried into another.

The industrial pages contain broad claims about collaborative operation and reduced need for cages. A robot’s safety features do not settle the behavior of a complete cell with tooling, sharp workpieces and connected machinery. Request the application-specific integration and safety documentation rather than turn a marketing statement into an operating assumption.

Public product descriptions also leave software access, update control and the boundary between autonomous operation and remote assistance unresolved for some configurations. Ask for a demonstration that identifies any operator intervention, then agree how intervention is recorded. This matters especially when comparing a humanoid demonstration with a production task that must run predictably across shifts.

07 / DecisionChoose the robot around a verifiable production boundary

NEURA is most useful to evaluate through a specific manipulation or transport problem. The next step should produce a supported configuration and an acceptance case that the intended operators understand.