Autonomous Self-Driving Lab for Robot-Assisted Scientific Experimentation

Mon, 01 Jun 2026 00:00:00 +0000

Project Mission

This project develops autonomous self-driving laboratory systems that integrate robotics, AI, sensing, and scientific instrumentation for end-to-end experimental execution.
We focus on building robotic workflows that can handle samples, transport materials across laboratory spaces, operate analytical instruments, monitor experimental conditions, and generate feedback for the next experimental cycle.

The long-term goal is to move beyond isolated laboratory automation toward a physically embodied discovery platform, where robotic agents can connect experimental planning, sample handling, measurement, analysis, and decision-making in a unified closed-loop system.

Scientific Motivation

Modern scientific experiments often depend on repetitive manual operations, fragile sample handling, instrument-specific procedures, and delayed analysis. These constraints limit throughput, reduce reproducibility, and make it difficult to run experiments continuously across long time horizons.

Self-driving laboratories provide a new framework for scientific discovery by combining AI-driven planning with robotic execution and automated characterization. In this project, we extend that idea from benchtop automation to a broader robotic laboratory architecture that includes dexterous manipulation, mobile sample logistics, real-time monitoring, and future integration with surgical and in vivo experimental systems.

Research Approach

Dexterous Manipulation and Touch Sensing

Touch-Aware Grasping: Developing robotic manipulation policies that use tactile feedback to handle fragile samples, containers, tools, and experimental materials.
Vision-Tactile Fusion: Combining camera-based perception with contact sensing to improve grasp stability, object recognition, and manipulation under uncertain laboratory conditions.
Low-Light Robotic Operation: Designing touch-centered control strategies for dark-lab or lights-out environments where visual sensing alone is insufficient.
Slip and Force Monitoring: Using tactile signals to detect unstable grasps, excessive force, surface contact changes, and potential sample damage.
Skill Learning for Lab Manipulation: Training robots to perform reusable laboratory skills such as picking, placing, uncapping, loading, aligning, and transferring samples.

Robotic Sample Transportation

Quadruped Sample Delivery: Exploring legged robots for transporting samples across laboratories, hallways, stairs, and uneven indoor environments.
Secure Payload Handling: Designing sample carriers that preserve orientation, reduce vibration, and maintain traceability during robotic transport.
Multi-Floor Laboratory Navigation: Planning robot routes that account for stairs, doors, elevators, restricted areas, and handoff points between workstations.
Chain-of-Custody Tracking: Connecting barcodes, RFID, or vision-based sample identification with automated experiment logs.
Human-Robot Lab Coordination: Developing safe interaction protocols for robots operating around researchers, instruments, and sensitive materials.

Automated Experiment Execution and Analysis

Instrument Integration: Connecting robotic arms, liquid handlers, incubators, microscopes, spectrometers, and environmental sensors into a coordinated experimental workflow.
Automated Sample Preparation: Standardizing preparation steps such as dispensing, mixing, incubation, transfer, and measurement setup.
Real-Time Experimental Monitoring: Tracking instrument status, sample condition, environmental variables, and abnormal events during autonomous operation.
AI-Based Data Interpretation: Applying machine learning to analyze imaging, spectroscopy, sensor streams, assay outputs, and experimental metadata.
Closed-Loop Optimization: Using experimental results to update hypotheses, refine protocols, and select the next set of experimental conditions.

System Design

The proposed system is structured as an embodied self-driving lab pipeline. An AI planning module selects experimental objectives and schedules tasks, robotic manipulators execute local operations, a quadruped robot transports samples between physical stations, and analytical instruments produce measurements that are automatically processed by data models.

The laboratory operating layer coordinates these components by managing task queues, robot availability, sample identity, instrument readiness, safety constraints, and data provenance. This architecture allows the platform to operate as a connected experimental ecosystem rather than a collection of independent automation devices.

Current Implementation

At the current stage, this project is structured around modular robotic capabilities rather than a fully autonomous dark lab. The initial focus is on developing robot manipulation policies, tactile sensing modules, sample transportation strategies, and software interfaces for experimental monitoring.

The near-term implementation will evaluate how dexterous manipulators and quadruped robots can coordinate with automated instruments for sample preparation, transport, analysis, and data logging. This stage is designed to establish robust physical execution before expanding toward more complex closed-loop scientific workflows.

Future Research Directions

Future work will extend the platform toward supervised robotic experimentation in biological and biomedical settings. One direction is to integrate surgical robotic systems with automated sample handling, imaging, physiological monitoring, and post-experiment analysis.

The long-term vision is to connect in vitro experiments, animal studies, and in vivo experimental workflows through a unified robotic pipeline. In this framework, surgical robots would support controlled intervention or sample acquisition, mobile robots would manage sample transfer, analytical systems would process measurements, and AI models would monitor outcomes under strict ethical, safety, and regulatory constraints.

This direction is not intended to replace scientific oversight. Instead, the goal is to develop a supervised autonomous experimentation system that improves precision, reproducibility, traceability, and continuity across the full experimental cycle.

In Vivo Automation | Research Lab