Automated Sample Collection Using a Quadruped Robot

Project Description

Amentum conducted a three-month R&D project for RAICo to develop a system to be used for autonomous sample collection through swabbing using the Boston Dynamics Spot® robot. We developed a system which provided:

1. Fully autonomous navigation.

2. Autonomous surface estimation and alignment procedures.

3. Autonomous swabbing using an arm-surface contact planner.

4. Integrated pressure measurement devices to qualify performance.

We provided RAICo with the finished system, all software and documentation for their long-term use, as well as experimental reporting produced during development, and consulting on pressure sensing technology.

We supported the client with a deployment of the full system at a stakeholder demonstration day, as well as hosting Factory Acceptance Testing and Training on our premises.

Services Description

We designed a multi-staged development process to ensure that the finished system would meet client requirements.

• Phase 1: Requirements Gathering & Design Proposal
• Phase 2: Autonomous Navigation Demonstration
• Phase 3: Swabbing Motion Trials & Data Capture
• Phase 4: Stakeholder Demonstration, FAT, Delivery & Handover

During Phase 1 we held a structured requirements capture workshop. In this workshop we elicited a set of requirements for the finished system to meet and addressed several areas of uncertainty that were remaining after submission of our initial proposal to determine the intent of the client. Using the developed set of requirements, we prepared a finalised design proposal along with a Requirements Traceability Matrix (RTM) which was updated and shared with the client periodically through the programme.

In Phase 2 we began development of the autonomous system by integrating a high-level controller and the Spot autonomous navigation stack. To control this system a web-based UI was created which gave significant mission planning, and data visualisation functionality with little computational overhead. We also developed a purpose-built alignment procedure to ensure that the Spot system could successfully swab at different locations on walls and floors. Once the system was capable of autonomous navigation and alignment, we ran tests and benchmarked the navigation performance. This was communicated in a technical note to the client.

In Phase 3 we developed a protocol for swabbing, using UKAEA’s swabbing tool, and integrated a pressure sensor to the system so that we could quantify swabbing performance with a reliable ground-truth. To do this we implemented the Arm Surface Contact Planner available in the Spot SDK, which uses a hybrid-motion force controller. We also provided data visualisation from the pressure sensors within the UI and prepared a suite of post-processed data which is extractable when using the UI. All this work was detailed in experimental reports which were communicated to the client.

In Phase 4 we installed the system on the client’s premises for a stakeholder demonstration and supported this demonstration. Between this activity and FAT, our development work focused on providing focused on providing tools for post-processing, viewing historic data and making some quality-of-life changes to the UI. The System was then subject to FAT, and the client attended our premises for a day of training with the system.

Benefits

This work provided the client with a step-change in their Spot capabilities. Particular benefits include:

• A first successful implementation of the arm surface contact planner for swabbing.
• All the development work was achieved in an 8-week window.
• Providing client support and impartial advice when they sought a change on the desired pressure sensing method.
• Accelerated development using the Spot SDK.