README ====== This file describes the dataset: data_extraction_by_system_deposit.csv 1. GENERAL INFORMATION ----------------------- 1.1 Dataset title Charted dataset: Robotic manipulation capabilities and end-effector technologies in chemistry laboratory automation -- a scoping review 1.2 Principal investigator / data collector Lupo Manes Department of Mechanical and Aerospace Engineering University of Liverpool Brownlow Hill, Liverpool, L69 3GX, United Kingdom Email: lupo.manes@liverpool.ac.uk 1.3 Associate / co-investigators Mario Gianni Department of Computer Science, University of Liverpool Brownlow Hill, Liverpool, L69 3BX, United Kingdom Email: mario.gianni@liverpool.ac.uk Paolo Paoletti (supervisor) Department of Mechanical and Aerospace Engineering, University of Liverpool Brownlow Hill, Liverpool, L69 3GX, United Kingdom Email: paoletti@liverpool.ac.uk 1.4 Contact person for questions Lupo Manes (lupo.manes@liverpool.ac.uk) 1.5 Date or date range of data collection Database searches were executed on 1 May 2026 (Web of Science Core Collection, Scopus, IEEE Xplore). This date is the end of the recall envelope: no record published after this date is included. No lower bound was placed on publication date; the 66 charted systems are drawn from 63 source publications spanning 1982-2026. Screening, data extraction and charting began on 2 May 2026 (the day following the search) and continued through 19 September 2026. 1.6 Geographic location of data collection Not applicable in the physical sense: this is a desk-based bibliographic review. The charted systems themselves are reported from laboratories worldwide; no single geographic location applies to the dataset. 1.7 Topic keywords Robotic manipulation; end-effector design; laboratory automation; chemistry automation; scoping review; gripper taxonomy; self-driving laboratories; PRISMA-ScR 1.8 Language English 1.9 Funding sources This review was conducted as part of a PhD scholarship funded by the Leverhulme Trust. 2. DATA AND FILE OVERVIEW -------------------------- 2.1 File description data_extraction_by_system_deposit.csv is the complete charted dataset underlying the review: one row per robotic system (66 rows), charted against an eight-descriptor manipulation taxonomy adapted from the ADePT framework. Every categorical descriptor value is accompanied by the verbatim source excerpt that supports it, and by a free-text charting note where relevant. 2.2 File format Plain-text, comma-separated values (CSV), UTF-8 encoded, with a header row. Fields containing commas, quotation marks or line breaks are quoted per RFC 4180. 2.3 Relationships among files This is a single, self-contained file: one row per charted system, with one column group per taxonomy descriptor (see Section 5). It is not part of a multi-file relational dataset. 2.4 File creation date 2026-09-20 2.5 Update / version history This is the first public version of the file. No prior versions have been deposited. 2.6 Related data not included in this deposit This file is a cleaned, public-facing export of a larger internal working spreadsheet used during the review. Two categories of column present in the internal working file were removed before deposit: - An internal row identifier generated by the screening platform (Rayyan), used only to cross-reference records between the screening tool and the extraction spreadsheet. - A "PDF checked" flag used internally to track verification progress during extraction (all 66 rows had been checked by the time of this deposit). The PRISMA-ScR screening-flow counts (numbers of records identified, screened, and excluded at each stage) are reported in the associated article and its Supplementary Information, rather than as a separate machine-readable file in this deposit. 3. SHARING AND ACCESS INFORMATION ----------------------------------- 3.1 Licence / restrictions on use This dataset is made available under a Creative Commons Attribution 4.0 International licence (CC-BY 4.0). You are free to share and adapt the material for any purpose, provided appropriate credit is given. Full licence text: https://creativecommons.org/licenses/by/4.0/ 3.2 Embargo status This dataset is deposited in the University of Liverpool Research Data Catalogue and is currently under embargo. It will be made publicly accessible on publication of the associated article. A human-readable copy of the same data is provided as Supplementary Information to the article in the interim. 3.3 Publications that cite or use this data Manes, L., Gianni, M., Paoletti, P. "Robotic manipulation capabilities and end-effector technologies in chemistry laboratory automation: a scoping review." Digital Discovery (Royal Society of Chemistry). 3.4 Other publicly accessible locations for the data A human-readable version of this dataset is included as Supplementary Information with the published article (see Section 3.3). 3.5 Recommended citation Manes, L. (2026). Charted dataset for: Robotic manipulation capabilities and end-effector technologies in chemistry laboratory automation -- a scoping review [Data set]. University of Liverpool Research Data Catalogue. 4. METHODOLOGICAL INFORMATION -------------------------------- 4.1 Data collection / generation methods Records were retrieved by searching Web of Science (Core Collection), Scopus, and IEEE Xplore on 1 May 2026, using Boolean queries built around a Population-Concept-Context (PCC) framework, executed as parallel title-field and abstract-field streams, plus three narrower queries targeting self-driving laboratories. The full, verbatim query strings for every database and stream are given in the article's Supplementary Information (Section S1.2). Records were de-duplicated and screened using Rayyan (https://www.rayyan.ai/): each record was screened independently and in duplicate by two of the three authors, with disagreements resolved by the third. The same duplicate-screening procedure was applied at full-text stage. Full eligibility criteria (Population, Concept, Context dimensions) are given in the Supplementary Information (Section S1.1). 4.2 Data processing methods Included sources were charted against an eight-descriptor manipulation taxonomy (Section 5 below), adapted from the ADePT framework, at the level of the individual robotic system rather than the publication: a single publication reporting more than one physically distinct system is split into separate rows, and multiple publications reporting the same system are merged into a single row (its contributing publications are listed in the source_publications and source_citations fields). Each descriptor was populated, in order of preference, from a source's main text, its supplementary material, and finally from figures or photographs of the setup; the evidence basis for tactile-sensing classifications specifically (reported vs. inferred) is recorded in the tactile_sensing_evidence_basis field. Every categorical value is accompanied by the verbatim excerpt that supports it. Full descriptor definitions and category lists are given in the Supplementary Information (Section S1.4). 4.3 Software and tools used - Rayyan (https://www.rayyan.ai/) -- de-duplication and title/abstract and full-text screening. - Python 3, with pandas, numpy and matplotlib -- data cleaning, cross-tabulation and figure generation. Specific package versions are not pinned in this deposit; contact the corresponding author if exact versions are required for reproduction. 4.4 Quality assurance procedures Every row was checked against its source PDF(s) at least once after initial extraction, and any discrepancy between a charted value and the source text was corrected, with the correction and its rationale logged in an internal audit trail (not included in this public export). Screening at both the title/abstract and full-text stages was performed independently and in duplicate, with disagreements adjudicated by the third, uninvolved author (Section 4.1). 4.5 Codes/symbols for missing or uncertain data See Section 5.4 (Codes for missing data) below. 4.6 People involved in collection, processing and analysis Screening and data charting: Lupo Manes, Mario Gianni, Paolo Paoletti (all three authors; see Section 1.2-1.3). Data curation, formal analysis, and preparation of this dataset for deposit: Lupo Manes. 5. DATA-SPECIFIC INFORMATION FOR: data_extraction_by_system_deposit.csv -------------------------------------------------------------------------- 5.1 Number of variables 31 columns. 5.2 Number of cases / rows 66 rows, one per charted robotic system. 5.3 Variable list Identification and bibliographic fields system_label Short human-readable label for the charted system (typically first author, year, and a brief descriptor). source_publications Short label(s) "Title (Year)" for every publication contributing to this row, semicolon-separated where a row merges more than one publication. source_citations Full formatted reference(s) for every contributing publication (authors, year, title, journal, volume/issue, DOI link), semicolon-separated, in the same order as source_publications. multi_robot_note Free-text note explaining how this row was scoped when its source publication reports more than one distinct robotic system. Blank where not applicable. same_system_cluster_note Free-text note explaining why two or more publications were judged to report the same physical system and merged into one row. Blank where not applicable. Taxonomy descriptor fields (repeated per descriptor; see Section 4.2) For each of the eight descriptors below, three columns are given: _value The extracted/verbatim value as charted. _quote The verbatim excerpt from the source that supports the charted value. May be blank where the value was inferred from a figure or photograph rather than text (see tactile_sensing_evidence_basis for the one descriptor where this is tracked explicitly). _category The standardised taxonomy category the value was mapped to (full category definitions in SI Section S1.4 of the associated article). mobility_value / _quote / _category Categories: Stationary; Mobile (mobile base); Mounted on rail. dof_value / _quote Degrees of freedom of the manipulator, charted as a numeric value. (No separate category column: DOF is charted directly as a number.) ee_design_value / _quote / _category End-effector form factor, independent of any modification applied. Categories: Parallel-jaw; Specialised; Soft; Not reported. ee_modification_value / _quote / _category Type of customisation applied to a baseline end-effector. Categories: None; Hardware; Sensing; Hardware + Sensing; Fully bespoke. tactile_sensing_value / _quote / _category Contact-sensing modality. Categories: None; Torque-based (arm); Torque-based (gripper); Force / pressure; High-resolution tactile. ("Not reported" is a valid category per the taxonomy but is unused in this dataset -- see Section S1.4 of the article.) tactile_sensing_evidence_basis Whether the tactile-sensing classification was explicitly reported by the source or inferred (e.g. from a platform datasheet or product page) -- see Section 4.2. task_types_value / _quote / _suggested_category Category/categories of manipulation task performed; a system may carry more than one tag, comma-separated within the cell. Categories: Sample/labware transport; Liquid handling; Instrument interaction; Assembly and setup; Solid handling; Multi-step workflow execution. env_customisation_value / _quote / _category Extent to which the laboratory environment was adapted for the robot. Categories: Bespoke workcell; Workstation-level fixtures; Labware-level adapters; Test setup; None. workflow_context_value / _quote / _category Whether the manipulation is embedded in an operational chemistry workflow or demonstrated as a standalone skill/benchmark. Categories: Embedded in chemistry workflow; Standalone skill / hardware benchmark. Limitations limitations_text Free-text summary of the hardware/manipulation limitation(s) reported for this system, if any. limitations_quote Verbatim excerpt(s) from the source supporting limitations_text. 5.4 Units of measurement No column in this file carries a physical unit directly (degrees of freedom, in dof_value, is a dimensionless count). Where a source's verbatim quote includes a measurement (e.g. torque, mass, a probe diameter), the unit is given as written in that quote and has not been standardised. 5.5 Codes/symbols for missing data "Not reported" Used within a _category field where the source gives too little detail to classify the descriptor. This is a deliberate, positive code, not a blank cell. "None" Used within ee_modification_category and tactile_sensing_category only, where the source positively establishes the absence of the feature (distinct from "Not reported" -- see Section S1.4 of the article). (empty cell) Used only in the free-text supporting fields (_quote, multi_robot_note, same_system_cluster_note) where no excerpt or note applies. No categorical (_category) field is ever left blank. 5.6 Specialised formats or abbreviations DOF Degrees of freedom PCC Population, Concept, Context (the review's eligibility framework) ADePT The pre-existing taxonomy framework this review's descriptors were adapted from PRISMA-ScR Preferred Reporting Items for Systematic reviews and Meta-Analyses extension for Scoping Reviews