Robots, data and agentic AI — as a structured academic program.
For deans and innovation directors: quadrupeds, wheel-legs and humanoids from our partner manufacturers, with technical support, an AI-robotics integration lab and research tracks for undergraduate, master's and doctoral students — in one agreement.
For universities, the same three layers become: certified equipment with support, an integration lab, and a research platform.
A robot without support, data and a curriculum is an expensive exhibit.
Universities in Peru and Brazil are acquiring quadrupeds, wheel-legs and humanoids at record pace. Three things decide whether the machine produces research or gathers dust: who keeps it running, how its data reaches the rest of the lab, and which students are structurally attached to it.
Support
Firmware, spares, calibration and a technician who answers — in Peru and Brazil, within 24–48 h.
Data
Robot sensors (LiDAR, cameras, IMU, thermal, gas) integrated with the lab's own sensors, software and industrial systems.
People
A track for each level — undergraduate, master's, doctorate — with defined deliverables, certification and co-supervision.
The AMARU University Program bundles the three into one agreement, so the innovation directorate signs once and the lab runs for years.
One brain. Every source. Real action.
An agentic AI platform that consolidates information from any source — robots, fixed sensors, operations systems — and acts online, in real time.
Perceive
Telemetry and sensor fusion: thermal, gas, acoustic, visual, LiDAR streams from every robot and fixed sensor.
Reason
Anomaly detection per asset, cross-source correlation, risk scoring and prediction.
Act
Schedules and re-plans rounds, raises alerts, dispatches service, writes reports.
For the university
The same NOC is offered as a sandbox: students write agents, connectors and models against real telemetry, and the best ones are deployed into live operations.
Quadrupeds and wheel-legs: we choose the machine for the task, not from the catalog.
B2
6 m/s · payload 40 kg+ · 5 h autonomy · IP67 · open SDK
Best cost-performance of the fleet; the natural first platform for a university lab.
AS2
Payload 16 kg · 2 h+ walking · ~25 km range · 5 m/s · IP66 · 25 cm step
Industrial-grade cross-roller bearings: high precision, high load.
X30
56 kg · payload 20 kg+ · −20 °C to +55 °C · IP67 · 45° slopes and stairs · 2.5–4 h
Fusion perception in smoke and darkness; proven in rescue drills.
Lynx S10
Under 20 kg · 3 h+ · IP66 · −20 °C to +55 °C · 4 cameras · front and rear LiDAR
Lightest and fastest; enters where the rest of the fleet does not fit.
From the teaching humanoid to the full-size outdoor platform.
G1
132 cm · 35 kg · 23 DoF (up to 43) · ~2 h · 2–3 kg arm payload · Jetson Orin-class compute · foldable, fits in a case
In the program: undergraduate coursework on locomotion, RL and manipulation; the Suzhou humanoid lab template.
H2
182 cm · ~70 kg · 31 DoF · ~3 h · ~7 kg arm payload (15 kg peak) · H2 Plus: Jetson Thor for vision-language-action models
In the program: master's and doctoral research; base of the NVIDIA Isaac GR00T research humanoid used by ETH, Stanford and UCSD.
DR01
170 cm · 80 kg · over 1.6 m/s · 15 kg payload · ~2 h · lightweight custom joints and limbs for complex terrain
In the program: doctoral research on whole-body control and terrain learning, on the same SDK family as X30 and Lynx.
DR02
IP66 full-body protection · −20 °C to +55 °C · modular quick-replace forearms, arms and legs · cargo transport and emergency response
In the program: joint pilots with AMARU's industrial clients — the humanoid enters the same NOC as the quadrupeds.
Two manufacturers, one representative, one support contract in Peru and Brazil.
Unitree
- Fleet in AMARU's line: B2 and AS2; G1 and H2 humanoids; Go2 for teaching labs
- Open-source stack: 15+ repositories, URDF/MJCF/USD models, unitree_rl_gym, SDK2, ROS packages
- Deep academic footprint: research humanoid with NVIDIA, embodied-AI industry colleges, joint research institute, RoboCup partner
Deep Robotics
- Fleet in AMARU's line: X30 and Lynx S10; DR01 and DR02 humanoids; Lite3 and Lynx M20 for education
- Open SDKs, GitHub resources, NVIDIA Jetson compute and ROS 2 across the education line
- Industrial references: power grids, converter stations, fire and rescue, ports and steel conveyors, Singapore SP Group
- Academic use: terrain-imagination and visual-navigation papers with Zhejiang and Hunan universities
Three pillars. One agreement. One lab that runs for years.
Technical support of the robot line
- Commissioning and safety protocol
- Operator training and certification
- Manufacturer warranty, firmware and spares in Peru and Brazil
- Field service 24–48 h and remote support
- Preventive maintenance and calibration
- SDK and developer access
AI robotics and data integration
- Sensor data pipelines (LiDAR, cameras, IMU, thermal, gas)
- Integration with external sensors and applications
- NOC sandbox with real industrial telemetry
- Digital twin and simulation environment
- Agent development and deployment
- Anonymized industrial datasets
Academic tracks by level
- Undergraduate: operator certification, integration bootcamp, capstone
- Master's: applied research and joint industry pilot
- Doctorate: original research on the NOC as testbed
- Competency ladder shared across levels
- Publications, theses and deployed agents as KPIs
- Faculty development and joint supervision
Governed by a joint steering committee; reviewed every semester against KPIs agreed on day one.
Everything needed to keep the fleet running, safely, in Peru and Brazil.
Commissioning on campus
Delivery, assembly, network setup, payload mounting and acceptance test with the lab team.
Safety protocol
Exclusion zones, speed limits, fail-safe stop and human override written into every lab mission.
Operator certification
AMARU-certified operators: faculty, technicians and students, with recertification every year.
Firmware and updates
Manufacturer firmware managed by AMARU; release notes and regression tests before each update.
Spares and field service
Spare parts stocked in Lima and São Paulo; technician on campus in 24–48 h; loan unit for extended repairs.
Remote support and SDK
Help desk with response SLAs, developer access to SDK and ROS 2 packages, quarterly preventive maintenance.
Applies to every robot line the university acquires through AMARU — from our partner manufacturers or integrated third-party units — under one support contract.
From robot sensors to external applications: the integration stack students build on.
Robot sensors
LiDAR · RGB and thermal cameras · IMU · joint encoders · gas · acoustic · GPS/RTK
Data layer
ROS 2 topics · MQTT · REST · OPC-UA · time-series store · georeferenced event log
AMARU NOC sandbox
Perceive → reason → act · anomaly models · agent runtime · digital twin (Omniverse) · anonymized industrial datasets
External applications and sensors
Lab bench sensors · campus IoT · SCADA/CMMS simulators · ML frameworks · dashboards · HPC clusters · partner systems
What students build
Sensor-fusion pipelines
Synchronize and fuse LiDAR, thermal and gas streams into one asset state.
Connectors
Bring an external sensor, program or database into the NOC as a data source.
Agents
Write perceive-reason-act agents that schedule missions and raise alerts.
Digital twins
Model the lab or a partner site and train policies in simulation first.
Dashboards and reports
Live KPIs, drift-vs-baseline views and audit-ready evidence packs.
Open standards throughout: ROS 2, MQTT, REST and OPC-UA. Nothing a student builds is locked to one manufacturer.
Six modules, each with a deliverable that runs on a real robot.
| Module | Skills | Deliverable | Level |
|---|---|---|---|
| M1 · Robot data foundations | ROS 2, SDK, telemetry, safety envelope | Mission runs with full telemetry logged to the data layer | Undergraduate |
| M2 · External sensor and application integration | MQTT/REST/OPC-UA connectors, time synchronization, data contracts | One external sensor or program integrated as a NOC data source | Undergraduate · Master's |
| M3 · Perception and sensor fusion | LiDAR + thermal + vision fusion, calibration, mapping | Fused asset state with confidence scoring | Master's |
| M4 · Anomaly detection and prediction | Baselines, drift models, failure-window estimation, CMMS hand-off | Predictive model validated on anonymized industrial data | Master's · Doctorate |
| M5 · Agentic operations | Agent runtime, planning, escalation policies, human override | Agent deployed to the NOC sandbox and reviewed for production | Master's · Doctorate |
| M6 · Multi-robot, humanoid and simulation | Digital twin, synthetic data, fleet coordination, humanoid whole-body control, sim-to-real | Policy trained in simulation and validated on quadruped or humanoid hardware | Doctorate |
Modules are delivered by AMARU engineers together with faculty, in blocks compatible with the academic calendar; each ends with a review in the NOC.
One competency ladder, three entry points.
Undergraduate
- Objective
- Certified operators and integration-ready engineers
- Format
- Bootcamp + semester course + capstone (TCC)
- Duration
- 1–2 semesters
- Deliverable
- One sensor or application integrated; mission portfolio
Master's
- Objective
- Fusion, prediction and agents validated on live data
- Format
- Research project + modules M2–M5 + joint pilot
- Duration
- 18–24 months
- Deliverable
- Dissertation, deployed model or agent, pilot report
Doctorate
- Objective
- New methods in coordination, agentic AI, humanoid control
- Format
- Co-supervised thesis + residency at the NOC
- Duration
- 36–48 months
- Deliverable
- Publications, open benchmark, production-grade agent
What each level masters — and where it hands over to the next.
| Competency | Undergraduate | Master's | Doctorate |
|---|---|---|---|
| Robot operation and safety | Certified operator; runs missions | Designs missions and safety envelopes | Defines fleet-level operating policies |
| Data integration | Connects one sensor or application | Builds multi-source pipelines and contracts | Architects integration for new sites and manufacturers |
| Perception and fusion | Uses fused outputs | Implements and calibrates fusion | Advances self-correcting perception |
| Prediction | Reads drift vs. baseline | Trains and validates predictive models | Proposes new prediction methods |
| Agentic AI | Runs and monitors agents | Designs and deploys agents to the sandbox | Researches agent architectures and governance |
| Simulation | Executes missions on the digital twin | Trains policies in simulation | Leads sim-to-real and multi-robot research |
| Output | Capstone, operator certificate | Dissertation, pilot report, deployed agent | Publications, benchmark, production agent |
Students can enter at any level; each track assumes the competencies of the previous one and closes with its certification: Certified Robot Operator, Integration Engineer, Research Fellow.
Clear roles, clear IP, one steering committee.
Agreement
Memorandum of understanding followed by a program agreement: equipment and support schedule, lab charter, academic annexes per track.
Steering committee
Innovation director, program coordinator, AMARU program lead and one industry partner; meets every semester, reviews KPIs and cohorts.
Intellectual property
University retains academic IP and publication rights; AMARU retains platform IP; jointly developed agents and models shared under pre-agreed terms.
Safety and compliance
AMARU safety standard applies to every lab mission; liability coverage and operator certification before the first step on campus.
Data
Industrial datasets anonymized and licensed for research; university data stays in the university; ethics review where human subjects are involved.
Industry link
Joint pilots and internships with AMARU clients in mining, plants, energy and security; demo days with invited companies.
Three ways to work with AMARU — every mode ends on the same program.
| Equipment supply + program | Lab-as-a-Service · recommended | Joint research pilot | |
|---|---|---|---|
| Who owns the fleet | The university | AMARU | Shared per project |
| Support | Pillar 1 tier of choice (Essential / Lab / Research Center) | Research Center tier included | Lab tier included |
| Integration lab and sandbox | Included with Lab tier and above | Included | Included |
| Academic tracks | Selected tracks | All three tracks | Master's and doctorate |
| Commercial model | Equipment purchase + annual support contract | Monthly subscription per robot, 24–36 months | Co-funded project (university, agency, AMARU) |
| Typical fit | Engineering schools with procurement budget | Universities preferring OPEX and fast start | Research centers with funded agendas |
Budgets are sized per campus after the lab assessment; public-procurement and funding-agency formats are supported.
From first meeting to first cohort in one semester.
Lab assessment
We walk the campus with you: space, network, safety, faculty interests and the robot line to acquire. Tracks and KPIs agreed in writing.
Commissioning and certification
Fleet delivered and commissioned, first operator cohort certified, sandbox and digital twin accounts opened.
Pilot cohort
Undergraduate course and first master's projects run; first integration delivered; steering committee reviews the KPI baseline.
Program at scale
All three tracks active, joint industry pilot under way, annual recertification and contract review.
We start with one lab and one cohort. We scale by track and by semester.
What we are asking of you today.
A lab assessment visit
2–3 weeks on campus with your faculty: space, network, safety, research interests and the robot line to acquire.
A memorandum of understanding
Equipment supply and support tier, sandbox access, and the academic tracks you want to open first.
A first cohort
One undergraduate course and two to four master's or doctoral projects in the next academic semester.
Innovation directorates that sign in this semester open the first cohort in the next one — with the fleet commissioned, operators certified and the sandbox live before day one of classes.
Program delivery is led by AMARU integration engineers and the NOC team, with faculty as co-instructors and co-supervisors. Who We Are →
In ten years, no person should enter where a machine can already go first.