# GEMM — Industrial robotics software

> Robots that work where people shouldn’t have to.

GEMM builds the software layer that enables robots to perceive, map and operate in complex industrial environments.

## We turn robots into industrial operators.

Commercial robots are increasingly capable. But operating reliably in the real world requires much more than hardware.

GEMM connects perception, mapping, control and progressive autonomy into one system designed around real industrial constraints.

**Perceive** → **Understand** → **Navigate** → **Operate**

## From data to real-world action.

One software layer connects what a robot senses to what an industrial team needs it to do.

### 01 — Perceive

Multi-sensor acquisition on the physical G1 platform.

**Status:** Physically validated

### 02 — Understand

Object detection and gauge-reading modules under integrated validation.

**Status:** Prototype

### 03 — Navigate

3D and 2D mapping are live; A-to-B navigation is validated in simulation.

**Status:** Validated in simulation

### 04 — Operate

VR hand and arm teleoperation works on G1; valve manipulation is next.

**Status:** Validated + in development

## What would you stop sending people to do?

GEMM is being built for repetitive, remote and hazardous work—so industrial teams can focus people where judgment matters most.

### Routine inspection

Repeatable rounds and consistent data capture across facilities.

**Status:** In development

### Hazardous environments

Reduce potential exposure to heat, gases, machinery and remote areas.

**Status:** Architecture direction

### Remote operation

Operate the robot from a safer location when human judgment is needed.

**Status:** Physically validated

### Industrial manipulation

Work toward interacting with valves, controls and human-built equipment.

**Status:** In development

**Building toward:** Thermal inspection · Gas detection · Acoustic monitoring · 3D mapping · Emergency assessment

## Autonomy where it works. Humans where they matter.

GEMM’s VR interface already enables low-latency hand and arm teleoperation on the physical G1. We use human control where judgment is essential while advancing toward higher levels of autonomy.

## The intelligence isn’t the robot.

GEMM is designed as a software layer around commercially available robotic platforms, so hardware can be selected for the environment instead of locking the system to one machine.

> Today’s development platforms are Unitree Go2 and G1 EDU. Portability beyond them is an architectural direction, not yet a commercial compatibility claim.

- **Go2:** Quadruped
- **G1 EDU:** Humanoid
- **What’s next?:** Many platforms, one intelligence

## A safer, more efficient industrial world.

We’re building the foundation for a future where robots can inspect, monitor and operate in demanding environments—with people always in control of what matters.

1. **Remote operation** — Physically validated
2. **Mapping & perception** — Physically validated
3. **Autonomous navigation** — Validated in simulation
4. **Autonomous inspection** — In development
5. **Physical interaction** — In development
6. **Industrial autonomy** — Architecture direction

## A research project becoming an industrial robotics company.

GEMM began in Buenos Aires with four students taking robot autonomy apart—from the sensor stream upward. The team first built control experiments around Go2, then moved into humanoid robotics with G1 EDU and a concrete industrial objective.

- **2026:** The team begins researching how far commercial robots can be pushed.
- **Go2:** Control experiments and early autonomy work.
- **G1 EDU:** Live mapping, a web dashboard and physical VR teleoperation.
- **Today:** Integrating the stack around real industrial constraints.

## A bigger tomorrow.

GEMM started with four students determined to push the limits of what’s possible with robotics. Today, we’re building technology that can make industrial work safer, more efficient and more human.

- **Guido Jacofsky:** Robotics & Teleoperation
- **Eitan Feldman:** Systems & AI
- **Martín Durán:** UX/UI Designer
- **Matías Tenenbaum:** Researcher

## Have a task in mind?

Tell us what you’d rather have a robot do. We’re looking to work with industrial teams on real environments, real constraints and real problems.

- Email: eitan@gemm.ar
- Website: https://www.gemm.ar

---

Spanish version: https://www.gemm.ar/es/index.md
llms.txt: https://www.gemm.ar/llms.txt