Location: London, UK ONLY. Office-based five days a week.
Employment: full-time, UK employment contract. Skilled Worker visa sponsorship and relocation support.
About the Company/Product:
- A global robotics company with hubs across North America and Europe, building robots for manufacturing, warehousing, and logistics.
- Its robots are already being piloted with industrial customers.
What You’ll Be Doing:
There are two open positions in the Control team in London. One is about the person controlling the robot remotely. The other is about the engineers who need hard evidence that a controller change actually improved anything. We’ll agree on the area and level after the interviews.
Teleoperation:
The goal is to give operators safe, precise, confident real-time control of the robot: to guide it through a task, recover it when something goes wrong, teach it, and improve how it behaves.
- Control loop: take raw, noisy operator input and turn it into smooth, stable motion that stays within what the robot can physically do and behaves the same way every time. This covers manipulation, moving around, and tasks that combine both.
- Operator hardware: connect VR controllers, gloves, haptics, motion capture, and custom control stations to the robot’s software.
- Operator experience: decide how controls map to robot motion, what the operator sees and gets back as feedback, and how they recover the robot after a failure. The aim is less mental strain and more confidence in the system.
- Help from the robot: build modes that combine what the operator intends with what the robot can do on its own, so control becomes more precise, safer, and easier.
- Safety: define safe operating limits and constraints, warnings, emergency and recovery procedures, and clear indication of the robot’s state.
- Infrastructure around the system: real-time streaming, calibration, session recording and replay, logs and debugging tools, operator training, and recording demonstrations clean enough to train models on.
- Proving it works: define and improve quality metrics, such as delay, reliability, observability, how often operators complete tasks, the quality of recorded demonstrations, and how the system fails. Push the system to its limits in simulation and on real robots, and make it more reliable based on what operators report and what breaks.
- Influence on the product: work with the controls, manipulation, hardware, and operations teams so that remote-control needs shape both the robot’s design and how it’s used on site.
Control Quality & Evaluation:
This is a software engineering role, not control design: you won’t be deriving a whole-body controller. Your job is to own how control quality is measured, on each robot and across the fleet. You’ll replace gut feeling with hard numbers that show whether a release beats the last one, robots behave alike, and a fleet is ready for a customer pilot.
- Fleet acceptance: create test suites that show whether a given robot, and the fleet overall, reaches the target level of control performance. Any team should be able to run them without a controls engineer on hand.
- Release gate in CI: build control assessment into the CI pipeline for both simulated and hardware-in-the-loop runs, with pass/fail criteria. Regressions should be flagged by the system itself rather than surfacing at a customer’s site.
- Repeatable test runs: develop automated test setups that assess each control release determ