Vento Autonomy

Perception and control for machines.

We connect perception, embedded software and control to investigate how machines act in the physical world. Our work spans a camera-guided mobile robot and controller firmware for a sit-to-stand walker, both at prototype scale.

Research & lab prototypes

Explore the current work Discuss a collaboration

Sensors, perception and a supervised control loop around a machine.

Current work

Current lab prototypes

Two prototypes let us investigate different parts of a machine system: navigation from camera input, and embedded control around a person’s inputs.

Camera-guided mobile robotLab prototype · Perception & navigation

Our mobile robot brings camera input, navigation and control into one small-scale system. It provides a way to study how observations inform movement decisions.

Research question: how should uncertainty in camera observations affect a movement decision?

Possible next experiment: compare observations, control commands and actual motion on a defined route as conditions change.

Discuss perception and navigation

Control firmware for a sit-to-stand walkerLab prototype · Embedded control

Our contribution is controller firmware for a sit-to-stand walker, linking inputs, control commands and machine state. The work also considers how that state is communicated to the person using it.

Research question: how should the controller respond when an input changes, arrives late or is absent?

Possible next experiment: use a bench setup with simulated inputs to compare expected commands and state changes, including stop conditions, before considering further trials.

Discuss an embedded-control problem

Research directions

Where this research could apply

Haulage, warehouse movement and repetitive site tasks raise related questions in perception and control. These are research directions, with requirements and validation to be established for each setting.

Haulage and fixed routes

Investigate perception and collision-warning capabilities for vehicles in controlled operating areas, with any further automation developed through staged evaluation.

Warehouse movement

Explore the sensing and control problems involved in repetitive material movement, including how a system responds around people and changing obstacles.

Repetitive site tasks

Study opportunities in tasks such as marking out and moving material, where a constrained first application could make the work measurable.

Research methods

How we test a machine system

A collaboration would start with a defined machine task, operating conditions and test scope. A focused prototype and measurements would help establish where the approach works and what needs further research.

Perception

Compare sensor observations with what is actually happening. Study missed detections, uncertainty and changing conditions before connecting an observation to an action.

Embedded control

Trace how inputs become commands and how the machine responds. Define operating limits and test the behaviour when information is delayed, incomplete or absent.

The operator interface

Make the machine’s state and intended action understandable. Define supervision and stop conditions alongside the control logic, then review the complete interaction.

Talk to Vento

What should the machine be able to do?

Share the task, the machine or hardware involved and the conditions it must work in. We can define a research question, prototype scope and evaluation plan.

Discuss an autonomy collaboration