Glossary¶
A reference guide for key terms and concepts in autonomous material handling.
Material Properties¶
- Yield Stress
- The minimum stress required to initiate flow in a material. Materials with high yield stress (like pastes and slurries) require more force to move.
- Rheology
- The science of material deformation and flow. Describes how materials respond to applied forces and stresses.
- Viscosity
- A measure of a fluid's resistance to flow. High viscosity materials (like honey) flow slowly; low viscosity materials (like water) flow easily.
- Flowability
- The ease with which a material can be moved, poured, or dispensed. Depends on particle size, shape, moisture content, and cohesion.
- Non-Newtonian Fluid
- A fluid whose viscosity changes with the applied shear rate. Examples: ketchup, blood, paint (shear-thinning), or Oobleck (shear-thickening).
- Cohesion
- The attractive forces between particles in a powder or semi-solid, affecting how they stick together and flow.
- Particle Size Distribution (PSD)
- The range and relative proportions of particle sizes in a powder. Affects flowability and packing density.
Handling Strategies¶
- Characterize, Then Act
- A strategy of measuring material properties first, then executing a pre-planned handling procedure based on those measurements.
- Iterative Optimization
- A strategy involving repeated cycles of dispense → measure → adjust to optimize process outcomes.
- Real-Time Correction
- A strategy that uses in-situ sensors to detect deviations during a process and corrects them on-the-fly.
- Feedforward Prediction
- A strategy that predicts the optimal process parameters before execution using prior knowledge or models.
- Simulation-Based Generalization
- A strategy of training models in simulation, then deploying them to handle real materials with different properties.
- Physical Reinvention
- A strategy of solving handling challenges by redesigning hardware (nozzles, feeders, grippers, etc.).
- Tactile Retrieval
- A strategy using touch-based sensing (force, texture, compliance) to manipulate objects without vision.
Hardware & Equipment¶
- Peristaltic Pump
- A pump that moves fluid by compressing and releasing a flexible tube in sequence. Good for viscous or particulate fluids.
- Syringe Pump
- A pump that dispenses fluid by pushing a plunger through a syringe. Precise but limited volume.
- Auger / Screw Feeder
- A rotating screw mechanism for handling powders and granules, providing controlled dispensing rates.
- Gripper
- A mechanical or pneumatic device for grasping and manipulating solid objects or partially formed items.
- Powder Hopper
- A container that feeds powder into a dispenser or processing unit, often with vibration or agitation.
- Rheometer
- An instrument that measures the flow and deformation properties of materials under controlled conditions.
AI & Control Concepts¶
- Feedback Control
- A control system that measures output and adjusts input to maintain desired performance (e.g., PID control).
- Feedforward Control
- A control system that predicts required adjustments based on known inputs, without measuring output.
- Cascade Control
- A hierarchical control structure where an outer controller sets targets for inner controllers.
- PID Control (Proportional-Integral-Derivative)
- A standard feedback control algorithm that adjusts output based on error (proportional), accumulated error (integral), and error rate (derivative).
- Machine Learning (ML)
- Algorithms that learn patterns from data to make predictions or decisions without being explicitly programmed.
- Regression
- An ML task that predicts continuous numerical values (e.g., predicting dispense volume from parameters).
- Classification
- An ML task that predicts categorical outcomes (e.g., predicting material type from sensor data).
- Digital Twin
- A virtual model of a physical system that simulates its behavior, used for testing and prediction.
- Discrete Element Method (DEM)
- A simulation technique for modeling particle interactions in powders and granular materials.
- Finite Element Method (FEM)
- A simulation technique for modeling continuous deformation in solids and semi-solids.
Process Metrics¶
- Precision
- The consistency or repeatability of a process. Low variance in repeated measurements or operations.
- Accuracy
- How close a measurement or result is to the true value or target.
- Throughput
- The volume or mass of material processed per unit time.
- Yield
- The percentage of material successfully processed or dispensed without waste or defects.
- Process Window
- The range of operating conditions (e.g., temperature, speed, parameters) within which a process works acceptably.
Cross-Cutting Terms¶
- Autonomous Formulation
- The automated design and synthesis of materials or products with minimal human intervention, often using feedback loops and optimization.
- Closed-Loop System
- A system with sensors that measure outcomes and adjust inputs to maintain control (feedback).
- Open-Loop System
- A system that executes pre-planned instructions without measuring or adjusting based on outcomes.
- Sensor Fusion
- Combining data from multiple sensors to improve understanding or control of a process.
- Payload
- The material being handled or dispensed in a robotic or automated system.
- Robustness
- The ability of a system to maintain performance despite variations in materials, conditions, or disturbances.
- Scalability
- The ease with which a process or strategy can be adapted to handle different material types, scales, or production rates.
Formulation SDL Terms (added with the V3 update)¶
- Self-Driving Laboratory (SDL)
- A laboratory that closes the loop between experiment design, robotic execution and data analysis, using results to select subsequent experiments with little human intervention [1].
- Premix container
- The vessel into which ingredients are dosed before it is carried to a mixer, processed and characterized. See Robot Classes.
- Object handling
- Moving the vessels and tools that hold or dispense materials (carrying containers, opening lids, capping, loading instruments), as opposed to material handling, which moves the matter itself.
- Hausner ratio
- Ratio of tapped to bulk density of a powder; ~1.00–1.11 indicates free-flowing and >1.4 very cohesive powder (Table 1 of the review).
- Sense–act sequence
- The three-verb title of each literature scenario (e.g. dispense → weigh → optimize) describing how a study senses a material and acts on it. See Literature Scenarios.
- Retrofit
- A modification of a purchased platform's end-effector, sensing or control software that unlocks a capability its manufacturer never sold. See Retrofits.
- Digital twin
- A persistent virtual counterpart of a specific system that shares its geometry, kinematics and device logic and is kept consistent with its measured state [89], [90].
- Fidelity level
- The amount of physics a simulation resolves: (1) workflow, (2) kinematics, (3) contact dynamics, (4) material physics. See Simulation & Digital Twins.
- DEM / MPM / SPH / CFD
- Discrete element method (powders), material point method (pastes and gels), smoothed-particle hydrodynamics and computational fluid dynamics (liquids).
- Sim-to-real gap
- The performance drop when a model or policy developed in simulation is deployed on hardware; for liquids, gels and powders it is set by the material model [128].
- ROS 2
- Robot Operating System 2, the middleware through which most research robots, simulators and planners exchange data [96].
- SiLA 2
- A laboratory standard defining typed device features with discovery, progress notifications and status subscriptions [167].
- OPC UA LADS
- Information models for laboratory and analytical devices built on industrial OPC UA [168].
- Orchestration software
- Software that runs dispensers, robots, instruments and twin as one system, ordering device operations and tracking resources and locations. See Orchestration.
- Functional layers (L1–L4)
- Device control (L1), device integration (L2), workflow orchestration (L3) and decision making (L4).
- Model Context Protocol (MCP)
- A protocol through which AI agents call tools; increasingly used to expose orchestration functions to agents [171].
- Surrogate model
- A model mapping composition and process parameters to measured properties, used in place of a lookup table.
- Acquisition function
- A rule that ranks candidate experiments using a surrogate's predictions and uncertainty, as in active learning and Bayesian optimization [134], [135].
- FAIR data
- Data that are findable, accessible, interoperable and reusable [133].
- Agentic system
- An LLM wrapped in a loop that plans, calls tools, observes results and revises its plan, as in Coscientist and ChemCrow [125], [127]. See LLMs & AI Agents.