Robotics & Automation for Food Production and Distribution

This maps the research literature of automation for food production and distribution — the field spanning the farm (harvesting, weeding, navigation, perception), the plant (processing, packaging, grading), and the network (supply chain, cold-chain logistics, delivery).

The shape of the field

The literature is large, applied, and dominated by perception and machine learning applied to agriculture rather than by mechanism or hardware. The single largest cluster is crop/fruit detection and phenotyping (deep-learning vision), followed by selective harvesting and by supply-chain/logistics work. "Smart Agriculture and AI" is by far the most common topic label — the field's centre of mass is software (models, IoT, decision systems), not robots.

Two distinct regimes are visible:

• High-volume, plateaued cores. Perception, harvesting, and supply-chain each produce 150–280 papers/year but grew slowly or flat over 2021–2024. These are mature, crowded areas — the vision-for-agriculture pipeline (YOLO-family detectors on crops) is now standard practice.
• Smaller, fast-accelerating frontiers. The steepest growth is in greenhouse / controlled-environment automation (~27%/yr), aquaculture robotics (~22%/yr), manipulation & grasping (~20%/yr), and field navigation / autonomous vehicles (~13%/yr). These are where the open problems — and the venture whitespace — currently sit.

Who is producing it

• China dominates by volume — ~1,900 papers (27% of the corpus have ≥1 Chinese author), roughly 2× the United States. Leading institutions: China Agricultural University, Jiangsu University, Northwest A&F, the Chinese Academy of Sciences, and Wageningen University (Netherlands), the long-standing Western centre of gravity for agri-food robotics.
• The venue landscape is applied and engineering-led. Computers and Electronics in Agriculture dominates, followed by broad-scope outlets (Sensors, IEEE Access, Agronomy, Remote Sensing). Robotics-native venues sit lower by volume — much of this work is published as applied agriculture rather than as robotics.

Most-cited work (2021–2026)

The most-cited papers skew toward reviews and AI/IoT framing pieces rather than robot-system papers — characteristic of a field still consolidating its vocabulary.

Fastest-rising recent work (2024+, by citation velocity)

The rising edge is dominated by generative-AI / foundation-model framing entering agriculture — GenAI in supply chains, LLM/foundation-model applications, "revolutionizing agriculture with AI." This is the newest inflection: the field is beginning to import the general-purpose-model wave.

Reading for an entrant from robotics + systems engineering

• The unsolved layer is physical, not perceptual. Vision is commoditized; dexterous, damage-free, cost-viable manipulation at field throughput is the recurring wall — the fastest-rising hardware theme.
• Controlled environments are where robotics economics work first. A structured environment collapses the perception-and-manipulation difficulty that open fields explode.
• The distribution half is under-roboticized relative to production. Supply-chain work skews toward traceability/blockchain/optimization software; physical automation of cold-chain, sorting, and last-mile food logistics is comparatively thin — a gap for someone bridging the data and physical layers.
• The foundation-model wave is arriving now (2024→). Whether it produces durable capability or a citation bubble is the live open question.