The Patent Landscape: Seven Jurisdictions, Four Technical Pillars

The swarm robotics patent landscape for warehouse logistics spans more than 30 active and inactive filings across South Korea, Japan, the United States, Israel, Germany, Singapore, and Italy. The dominant technical themes cluster around four pillars: distributed task allocation and scheduling, real-time swarm orchestration without fixed central controllers, multi-robot motion planning in shared workspaces, and dynamic obstacle detection for autonomous navigation. Together, these filings define the technological substrate enabling decentralized swarm coordination in high-density logistics environments — and they reveal a clear trajectory from rule-based fleet management toward AI-driven, self-organizing systems.

The most prolific assignees in the dataset include NAVIGATE LTD. (Israel, three filings on autonomous swarm orchestration), ABL IP HOLDING LLC (US, three filings on mechatronic agent swarms), REALTIME ROBOTICS, INC. (Japan, three filings on shared-workspace motion planning), Tata Consultancy Services (Japan, two filings on distributed knowledge and control solutions), and Korea Defense Science and Technology Institute (Korea, three filings on autonomous multi-robot collaboration). While not all patents are exclusively warehouse-focused, they collectively define the mechanisms by which swarm systems achieve coordination without a master controller.

Distributed Task Allocation and Semantic Knowledge Sharing

Decentralized warehouse swarms solve the task assignment problem by distributing knowledge to the robots themselves, rather than routing all decisions through a central server. Tata Consultancy Services’ 2024 patent on cooperative task execution describes a semantic knowledge base distributed among the robots in the fleet: each robot triggers self-queries and external queries to gather task-specific parameters, then uses the aggregated data to execute tasks assigned to the robotic group. Critically, this design ensures that task execution is unaffected even when the connection between robots and any centralized database becomes unstable — a condition that is common in large warehouse environments with dense metal shelving and high RF interference.

The software infrastructure for generating and deploying such distributed control logic is addressed by two Tata Consultancy Services filings from 2020 and 2023. These describe a knowledge bus architecture that includes repositories for robot capabilities and contextual problem spaces, a solution synthesizer that converts specifications into deployable designs, and a simulation-based solution verifier. This pipeline enables the rapid instantiation of decentralized control logic tailored to specific warehouse floor configurations, reducing dependence on hand-coded centralized planners — an important practical advantage as warehouse layouts change seasonally or in response to demand shifts.

Earlier foundational work from Daegu Gyeongbuk Institute of Science and Technology (2010 and 2011) introduced hardware-resource-aware task distribution: a robot task control module that selects the most capable robot for each generated task by collecting hardware and software resource information — including battery state and device availability — from each unit in the fleet. Although these patents have since lapsed, they established the architectural patterns that are now mature in current filings. The lineage from explicit hardware inventory to semantic knowledge distribution to AI-driven matching represents the central evolutionary arc of decentralized task allocation in swarm robotics, as documented by organizations including IEEE in its robotics standards literature.

Real-Time Swarm Orchestration Without Fixed Central Control

The most direct challenge to conventional fleet management is achieving globally coherent behavior across hundreds of robots without any persistent inter-robot coordination channel. NAVIGATE LTD.’s trilogy of patents — filed in 2023 and 2024 — articulates a coordination paradigm where the primary optimization objective is the maximization of short-term and near-real-time swarm objectives, not the fulfillment of any single robot’s goal. This is a critical philosophical and technical departure from classical centralized fleet management systems.

“Rather than routing all decisions through a single controller, each swarm member participates in a non-coordinated but globally coherent strategy, using geo-positioning data for navigation, mapping, and real-time state updates.”

The NAVIGATE filings explicitly acknowledge that prior art — including US10937324 — focused on individual drone coordination with other swarm members, while their invention pivots the focus to collective swarm-level outcomes achieved without persistent inter-robot coordination. This decoupling of individual behavior from collective result is the core property enabling scalability in warehouse environments where robot counts can reach into the hundreds and communication bandwidth is a finite resource. The approach has direct parallels with stigmergic coordination observed in natural swarms, a principle well-documented in research published by Nature.

ABL IP HOLDING LLC’s mechatronic agent swarm patents (2019 and 2020) describe a complementary architecture: each robotic agent carries its own propulsion, communication system, and power supply — a self-sufficient unit capable of receiving operating instructions via peer communication systems rather than a central server. The associated methods patents detail how a lightweight orchestration controller broadcasts reconfiguration instructions over a communication network, causing agents to holonomically reposition themselves without any agent requiring complete knowledge of the entire swarm state. Although the ABL patents target luminaire positioning rather than logistics per se, the holonomic multi-agent repositioning architecture directly parallels the motion management needs of warehouse AMR swarms.

Multi-Robot Motion Planning and Collision Avoidance in Shared Workspaces

Warehouse floors are shared operational spaces where the primary engineering risk is inter-robot collision — and where a centralized traffic controller becomes a bottleneck as fleet density increases. REALTIME ROBOTICS, INC. addresses this with two pivotal filings from 2022 and 2023: these systems represent the planned motion of each robot as a dynamic obstacle for all other robots simultaneously executing motion plans. Edges in the motion plan graph are assigned cost values based partly on collision assessments, and completed motion obstacles are pruned from the graph in real time. Motion plan requests are queued, and robots can be skipped or held in response to blocked conditions — allowing the swarm to self-regulate traffic without a central traffic controller adjudicating every movement.

A companion patent from REALTIME ROBOTICS, INC. (2022) extends this by storing discretized voxel representations of each robot’s swept volume and surrounding environment directly on the processor. At runtime, robots dynamically switch between stored motion planning graphs as their operational context changes, enabling low-latency path replanning without external computation offload — a property essential in warehouse environments where conveyor positions, pallet placements, and human worker locations change continuously.

Interference avoidance between discrete robot arms operating in overlapping zones is addressed by DENSO WAVE INC.’s 2021 patent, which defines occupancy zones for each movable part at its target position, identifies interference areas from estimated position and orientation, and uses these zones to coordinate arm movements without requiring a central arbitration layer. Sogang University’s 2024 filing takes a complementary approach for manipulation tasks: it generates independent object relocation work plans for each robot, compiles them into a collaboration plan that maximizes work shift numbers, and then sequences unit tasks for the shortest overall execution time while preventing inter-robot collision — a complete decentralized planning pipeline for logistics contexts. Standards bodies such as ISO have begun formalizing safety requirements for exactly these kinds of shared-workspace robot deployments.

Earlier geometric work from Kyung Hee University (2012) uses Voronoi points to enable distributed robots to establish optimal routes with minimal communication overhead. This design principle remains highly relevant in large-scale warehouse deployments where minimizing wireless traffic reduces latency and collision risk — a constraint that has only grown more pressing as AMR fleet densities have increased.

Autonomous Navigation, Obstacle Discrimination, and AI-Driven Robot Matching

For a swarm to operate without centralized control, each individual robot must navigate reliably in a dynamically changing physical environment — and must do so without corrupting the shared map that the entire fleet depends on. Tweeny Inc.’s 2023 patent addresses one of the most persistent problems in autonomous warehouse navigation: distinguishing transient objects (other robots, workers, misplaced pallets) from permanent environmental features. The system uses a LiDAR unit coupled with a node creation module, loop constraint calculation, depth image generation, and a weight adjustment module to build and continuously update a map that correctly classifies obstacles as temporary or fixed.

This capability is foundational for swarm systems. If robots cannot distinguish temporary obstacles from walls, they may permanently reroute around objects that will soon be cleared, creating system-wide inefficiencies that cascade across the entire fleet. The LiDAR-based discrimination approach represents a significant advance over earlier proximity-sensor systems, which lacked the spatial resolution to reliably classify obstacle permanence in cluttered warehouse aisles. Research organizations including WIPO have tracked the rapid growth of LiDAR-based navigation patents as a key indicator of autonomous logistics maturity.

Real-time cooperative robot control using AI-based matching is described in Punjin Co., Ltd.’s 2025 patent. This system uses a robot collaborative matching AI model to dynamically assign at least two robots from a fleet to perform a single mission collaboratively, controlling them in real time and releasing their pairing upon mission completion. The use of a learned AI model for matching — rather than a fixed rule-based dispatcher — enables the system to adapt to varying fleet sizes, robot states, and mission types without reprogramming a central scheduler. This is a qualitative shift: the coordination logic is no longer a static artifact but a continuously improving model.

Autonomous collaboration in multi-robot systems is further addressed by Korea Defense Science and Technology Institute’s 2025 filings. One patent computes expected reward values for each robot based on environmental information, control means, and situational decisions, then calculates a situation response degree across the multi-robot system — a reward-based autonomous decision mechanism that mirrors reinforcement learning principles and allows the swarm to self-organize responses to unexpected events. A companion filing enables robots to compete for situation-responsive control means when events occur, drawing decisions from a predefined pool — a mechanism that distributes authority without eliminating structure.

An early autonomous multi-robot cooperative baseline was established by Japanese inventor Toshio Fukuda’s 2009 patent, where each robot independently calculates an evaluation function combining distance to target, area of work zone, and dispersion distance relative to other robots already working nearby. This evaluation-function-based local decision making, requiring no global coordinator, represents one of the earliest patent-level formalizations of decentralized swarm logistics behavior — and the lineage from this approach to today’s AI-driven reward computation is direct and traceable.

Key Players and the Shift Toward AI-Driven Swarm Architecture

Analysis of the dataset reveals a clear shift from early (pre-2015) task assignment architectures requiring explicit hardware inventories toward post-2020 AI-driven, reward-based, and semantically distributed systems capable of adapting dynamically to real-time warehouse conditions without a global controller. Each of the five primary innovation clusters reflects a distinct layer of the decentralized coordination stack.

NAVIGATE LTD. (Israel) is the most focused swarm-specific filer, with three active patents centered on real-time autonomous swarm orchestration prioritizing collective objectives. Their approach explicitly rejects individual-robot coordination in favor of swarm-level outcome maximization — a philosophically decentralized stance with significant implications for warehouse scale-out.

REALTIME ROBOTICS, INC. (Japan/US) holds three active patents covering graph-based shared-workspace motion planning with real-time collision assessment. Their approach of treating each robot’s planned motion as a dynamic obstacle for all others — and pruning completed obstacles from the graph — represents the state of the art in decentralized collision avoidance for dense robot deployments, and is consistent with trajectory-space planning principles recognized by bodies such as the IEEE Robotics and Automation Society.

Tata Consultancy Services (Japan) contributes two active patents addressing the infrastructure layer: a distributed semantic knowledge base enabling decentralized task execution, and a knowledge bus architecture for generating and deploying robotic control solutions. These filings target the software fabric underlying any decentralized swarm system. Korea Defense Science and Technology Institute (Korea) holds three active patents on autonomous multi-robot collaboration using reward computation, event-driven decision pools, and collaborative control means selection — mechanisms directly analogous to the decentralized decision architectures required in warehouse logistics swarms, despite their defense-sector framing.

The overall trend is unambiguous: the patent record documents a systematic migration away from centralized control toward architectures in which intelligence is distributed across the fleet itself — making the swarm more resilient, more scalable, and more capable of operating in the unpredictable physical environments that define real warehouse floors. For IP professionals and robotics engineers tracking this space, the PatSnap patent search platform and PatSnap technology landscape tools provide direct access to the full filing histories of each assignee discussed here.