Platforms, sensors and models
Evaluate quadruped, humanoid or drone hardware against the task. Define sensors, model behaviour and on-board computing requirements within the manufacturer’s limits.
Our offerings
03 / AXENTYR Robotics
Put robots to work.
Connect sensors, AI models and operators to a defined mission on a quadruped, humanoid or drone, with links to the client’s existing systems.
Who it is for
Industrial, logistics, operations and site security managers considering missions such as inspection, internal transport, observation or assistance in a suitable physical environment.
Development status
In development. Mission software, hardware access and support arrangements are being defined by project; no equipment catalogue or deployment is announced.
The proposition
Give the hardware a mission and a place in your workflow.
Define what the robot observes, how a model interprets it, what reaches the operator and how approved results enter the client’s tools. Mission intelligence may run on board or use an agreed connection, according to latency and connectivity constraints. Manufacturer controls and physical safety remain independent of generative AI.
Start from an inspection, assistance or observation task. Assess a suitable quadruped, humanoid or drone, then define the sensors, models, on-board computing and operator controls needed for that mission. Hardware sourcing is evaluated for each project rather than treated as an available catalogue.
Explore the Platform architecture A shared target architecture, with no required bundle of offerings.
Scope
A focused scope.
Mission intelligence and operator control
Specify mission stages, observations, uncertainty and operator review, including what happens when a model cannot interpret a situation or connectivity is interrupted.
Existing systems and operating support
Connect approved observations and mission reports to client tools. Define training, software updates, hardware maintenance, support and intervention responsibilities.
How the work is structured
From the task
to a reviewed result.
An engagement starts from your objective and defines the inputs, work and validation needed to address it.
- 01
Define the mission and environment
State the task, site, permitted operating area and evaluation criteria. Identify the people involved and the manufacturer or regulatory constraints relevant to the platform.
- 02
Assess hardware, sensors and computing
Evaluate candidate quadruped, humanoid or drone equipment and project-specific sourcing. Select sensor inputs, AI models and on-board or connected computing for the mission.
- 03
Integrate mission logic and supervision
Define mission stages, operator controls, client interfaces and behaviour when data is uncertain or connectivity is lost. Keep physical safety controls separate from generative AI.
- 04
Evaluate with operators before acceptance
Test the scoped mission under agreed conditions, review observations and reports, and define training, maintenance and support before any operational acceptance.
Questions in context
See how a project
could be scoped.
Illustrative situations. These examples describe a question, the inputs and a result to review; they are not accounts of delivered work.
Quadruped
Inspect defined areas of a facility
- Question
- How could a quadruped gather useful observations along an approved inspection route and send them to the maintenance team?
- Inputs
- A permitted route, manufacturer-approved sensor data, the site’s inspection criteria, selected perception models and authorised interfaces to maintenance records. On-board resources and connectivity are assessed for the task.
- Output
- A mission report linking observations to location and time, candidate anomalies with uncertainty, incomplete checks and information for the existing maintenance workflow.
- Human review
- An operator validates the observations and requests further inspection where needed. Route permissions and independent physical safety controls remain with the responsible site team and equipment configuration.
Humanoid
Assess assistance between workstations
- Question
- Could a humanoid assist with a defined task sequence while operators remain responsible for changes and exceptions?
- Inputs
- Approved task instructions, workstation constraints, permitted sensor inputs, the selected hardware’s supported actions, model requirements and interfaces to the client’s task system.
- Output
- A mission sequence and task-status report, with observations requiring operator review, blocked steps and explicit handover points for tasks outside the assessed scope.
- Human review
- The responsible operator confirms the task interpretation and handles exceptions. Hardware suitability and physical safety require their own assessment; generative AI does not override safety functions or authorise movement.
Drone
Observe an authorised site area
- Question
- How could a drone mission turn permitted observations into a report that site operators can verify and use?
- Inputs
- An authorised mission area and flight plan, approved sensors and perception models, operator procedures, connectivity constraints and the agreed link to the client’s reporting tools.
- Output
- Time- and location-referenced observations, candidate findings with uncertainty, gaps caused by incomplete capture and a report for the existing site workflow.
- Human review
- A qualified responsible operator checks mission permissions and verifies the findings. Flight, navigation, emergency and physical safety controls operate independently of the generative interpretation layer.
Deliverables
A defined scope.
Useful deliverables.
- A mission specification and hardware, sensor and model selection criteria.
- An on-board/connected architecture with operator controls and integration interfaces.
- A reviewed mission report with observations, uncertainty and task status.
- A site evaluation, acceptance plan, training and maintenance responsibilities.
A worked example
A mission plan with a visible stopping point.
A fictional mission layout connects equipment, operator review and a simulated loss of connectivity. The mission pauses while independent safety systems remain in control.
Synthetic data. Every source, asset, state and decision below belongs to this example.
Mission
Inspect the agreed checkpoints, then return to the dock.
Fictional layout. The route and states illustrate mission supervision, not an available fleet.
Mission layout
A bounded inspection route.
- Planned route
- Restricted area excluded
- Q paused at the checkpoint
Equipment states
Read the task, link and operator.
Quadruped Mission paused
Inspection unit Q
- Task
- Checkpoint inspection route
- Connectivity
- Simulated link lost
- Operator
- Site operator review required
Humanoid Task not started
Workcell unit H
- Task
- Assistance inside the workcell
- Connectivity
- Simulated link available
- Operator
- Assigned workcell operator
Drone On the ground
Observation unit D
- Task
- Observation plan awaiting review
- Connectivity
- Simulated link available
- Operator
- Pilot authorisation required
Simulated incident
The task pauses when the link is lost.
Read the simulated connection incident
- Link loss recorded The example reports lost connectivity for inspection unit Q.
- Mission paused Further task steps are withheld. The last displayed position is a record, not a live location.
- Operator review requested Confirm connectivity, equipment condition and a safe route before considering a restart.
- Restart not authorised No instruction is sent. Any real restart requires the assigned operator and the manufacturer’s procedures.
Independent safety boundary
Generative AI does not own the safety system.
Manufacturer navigation, protective functions and equipment safety systems remain independent of generative AI.
A mission proposal cannot bypass operating limits, restricted areas or an operator’s authority. The selected equipment’s safety response to link loss must be validated for the real site.
Illustration only — no robot connection or commandHardware references
Different forms.
Defined missions.
Reference hardware — third-party examples. No partnership or availability implied. These photographs illustrate hardware categories for discussion; they are not an AXENTYR product catalogue.
Quadruped
Spot
Boston Dynamics quadruped, photographed on a forest path.
Photo: Jonte.
Humanoid
Robonaut 2
NASA and General Motors research robot, shown with an extended arm.
Photo: NASA / Robert Markowitz.
Drone
Drone with LiDAR
A quadcopter with a LiDAR payload, photographed on the ground.
Photo: Jonte.
Photographs resized and converted for web display, without cropping. The two photographs by Jonte remain available under CC BY-SA 4.0; the NASA photograph is public domain. Hardware selection and operating conditions would be assessed for each engagement.
Engagement options
A scope of its own.
A contract of its own.
Each offering has its own contract. Select the scope, access model and support arrangements that fit your needs.
Pricing structure
- Site preparation, hardware access and deployment integration, priced for the selected use case.
- Recurring software, hardware maintenance and support, with coverage and responsibilities specified.
- Replacement hardware and interventions with explicit cost and authorisation terms.
Before an engagement
Conditions to agree.
Prerequisites
- A defined mission, site constraints, operating permissions and evaluation criteria.
- A hardware feasibility assessment and project-specific sourcing conditions.
- Permitted sensor data, model requirements, computing and connectivity limits.
- A named operator, independent physical safety controls and acceptance responsibilities.
- Agreed interfaces, training, maintenance, support and replacement terms.
Operating responsibilities
- Hardware selection depends on availability, site suitability and the manufacturer’s operating requirements.
- Navigation, emergency controls and physical safety must remain independent of generative AI and within the manufacturer’s operating limits.
- Manufacturing, purchase, rental and operated-service models remain undecided; project-specific hardware sourcing is not a supply guarantee.
- Agree responsibilities for software, hardware, maintenance, replacement and interventions before deployment.
Independent offerings
Explore the other offerings.
AXENTYR Robotics can be considered on its own. A combination is optional and starts with your needs.
01
AXENTYR Intelligence
Turn authorised sources into a clear view of the situation, its relationships, conflicting evidence and the options to examine.
Explore this offering 02AXENTYR Security
Make assets, dependencies and coverage visible, then qualify alerts and coordinate the investigation of a cyber or operational incident.
Explore this offeringUseful questions
Questions about Robotics.
Does AXENTYR manufacture or sell robots?
Manufacturing and purchase, rental or operated-service models remain undecided. The role described here is mission intelligence and integration. Hardware sourcing is assessed for each project according to suitability, availability and supplier terms; this is not an immediate supply promise.
What does robot mission intelligence include?
Mission planning, sensor inputs, AI models, on-board or connected computing, operator review and links to client tools. Navigation, emergency controls and physical safety must remain independent of generative AI and respect the manufacturer’s operating limits.
How would hardware be selected for a site?
Start with the task and environment, then consider a suitable quadruped, humanoid or drone. Assess sensors, model and computing needs, operator controls, manufacturer requirements and project-specific sourcing before defining acceptance and support.
Can Robotics be contracted on its own?
Yes. Robotics has its own scope and contract, independently of Intelligence or Security. It can be designed around the client’s existing tools; the interfaces, hardware access, supervision, training and maintenance responsibilities must be agreed for the mission.
AXENTYR Robotics
Define the mission. Assess the integration.
Describe the task, site and platform you are considering. Scope the hardware assessment, mission intelligence, operator workflow and interfaces as one robotics project.
Scope a robotics project