What is Vertex? The leaderless consensus engine for real-time coordination →
HACKATHON

The Vertex Swarm Challenge 2026

Coordinating Robots, Drones, AI Agents, and IoT Fleets Beyond the Cloud

USD 27,000+ worth of prizes and grants, including accelerator tickets
Register Now → View Tracks

Event Timeline

Submissions open in 0 days

Phase 1

Submission

2026 / 03 / 17 01:30 UTC
Phase 2

Deadline

2026 / 04 / 06 13:30 UTC
Introduction

The missing TCP/IP for swarms

The Vertex Swarm Challenge 2026 is a global engineering challenge for developers, AI builders, researchers, and system architects looking to define the next era of peer‑to‑peer coordination for machines.

We are not interested in how a single robot moves in a perfect lab, or how a single LLM answers a prompt in isolation. We are here for the moment when dozens of heterogeneous drones, robots, and AI agents discover each other, negotiate roles, and execute as one swarm in the chaotic real world.

True autonomy requires machines talking directly to machines. Whether you are orchestrating a fleet of Autonomous Mobile Robots (AMRs) or a decentralized mesh of AI agents, Vertex 2.0 is the missing coordination layer, the TCP/IP for swarms.

You will use Vertex to make machines discover each other, share state, and coordinate decisions locally, without relying on a single orchestrator, centralized cloud, or vendor middleware.

Coordinate

Multi-vendor robots, drones, and AI agents that auto-discover, share state, and cooperate as one swarm. Different brands. Same swarm.

Automate

Fleets that negotiate routes, hand off tasks, and self-heal when units drop or networks degrade. No single point of failure. No manual intervention.

Secure

Safety signals that propagate instantly through the mesh. One node detects a fault → the entire fleet freezes in milliseconds.

Your Challenge

Eliminate the middleman.

Whether you are a solo builder making your first peer-to-peer handshake or a team architecting a multi-vendor robotic swarm, the mission is the same.

Low Barrier, High Ceiling Get your first resilient mesh running in under an hour. Get 10+ agents coordinating as a resilient swarm by the end of the challenge.
Systems Over Demos We reward coordination depth, reliability, low latency coordination, and real-world robustness over simple demos.

Prizes & Grants

If you build something real, we will back you with grants, credits, and accelerator support. Totaling USD $27,000+.

$6,250
Cash & Amazon Gift Cards
$12,000
Tashi Network Utility Grant / Tashi Token
$9,000
Accelerator Ticket

The Tracks at a Glance

Choose your battleground: from a quick environment test to open-ended swarm creativity, mission-critical robotics, or leaderless AI Agent economies.

Warm Up Community Badge

The Stateful Handshake

Prove your P2P connection with discovery, heartbeats, and shared state.

Before you build a massive autonomous fleet, you need to prove the fundamentals of peer-to-peer coordination. This warm-up moves beyond a simple "ping" and allows you to prove that two agents can discover each other, stay in sync, and recover from failure. This warm‑up should take less than an hour for most developers.

What participants do
  • Spin up: Launch two Vertex nodes in local terminal windows or simulation containers.
  • Discover & Handshake: Nodes find each other via Vertex P2P discovery. Establish a session/channel and exchange a signed "hello" payload.
  • Pulse Check: Send periodic heartbeats both ways to keep the connection alive.
  • Sync State: Maintain a tiny replicated JSON state across both nodes containing: peer_id, last_seen_ms, role, status.
  • Trigger Action: Agent A toggles its role. Agent B must mirror this state change in <1 second.
Acceptance Criteria
  • A short screen recording or terminal log snippet showing discovery, handshake, and active heartbeats (30-60 seconds).
  • State replication from Agent A to Agent B.
  • Failure Injection: Kill one node for 10 seconds to prove the peer is marked "stale," then bring it back online to show the connection and state automatically resuming.
Reward

Drop your proof into the Discord #shipping-log channel to claim your "Stateful Handshake" community badge and unlock your environment for the main tracks!

Track 1 Open Track

Ghost in the Machine

Exploring New Frontiers in Decentralized Swarm Coordination

This track is for teams applying Vertex's peer-to-peer coordination model to novel, domain-specific, or highly creative multi-agent systems. Your objective is to demonstrate direct coordination between multiple heterogeneous agents, whether they are physical robots, drones, or purely digital AI agents.

Teams are completely free to define their own use case, provided it showcases peer discovery, state sharing, and coordinated decision-making without a central broker.

While hybrid (Cloud + Edge) setups are permitted, submissions that demonstrate continued operation during partial failures, degraded connectivity, or a complete loss of centralized cloud services will score significantly higher for technical robustness.

Example Scenarios
  • Cooperative Perception: 3+ rovers sharing state so one instantly reroutes when another detects a blocked path.
  • Collective Decision-Making: 10+ heterogeneous agents reaching unified BFT consensus without a central server.
  • Decentralized Airspace Control: Drones negotiating flight corridors without a central ATC.
  • Smart Warehousing: 3-10 mixed legacy/modern AMRs coordinating tasks peer-to-peer.
  • Safety Mechanisms: Low-latency mesh fault detection where one failure freezes the whole swarm.
Judging Criteria (How to Win)
  • Innovation & Novelty: Creative multi-agent problem solving.
  • Decentralized Logic: State sharing and coordination without a central broker.
  • Robustness: System survives partial network failures or loss of cloud.
  • Developer Clarity: Clean, reproducible repository.
Track 2

Search & Rescue Swarms

Coordinating Multi-Robot Missions in Blackout Environments

Design and demonstrate a multi-robot or multi-drone swarm that can discover peers, share state, and coordinate search-and-rescue behaviors in environments where network connectivity is unreliable or completely unavailable.

Teams must use Vertex 2.0 to coordinate a swarm of at least 5 simulated robots and/or drones (in Gazebo or Webots), operating in a communications-degraded or blackout scenario. The focus is strictly on peer-to-peer coordination logic, not perception accuracy, mapping quality, or SLAM performance.

The Tashi Integration Pillars
  • Mesh Resilience (Vertex 2.0): Local swarm maintains BFT consensus despite high latency or packet loss.
  • Distributed State (FoxMQ): Shared "world map" array securely propagated across the fleet.
Example Scenarios
  • The Dynamic Daisy Chain: Drones act as P2P network relays to pass state up a structurally complex environment (like a tunnel).
  • The "Fallen Comrade": Instantly detect a simulated failure and re-negotiate unsearched territory among remaining nodes.
  • The Blind Handoff (Air-to-Ground): Aerial drone coordinates a rescue target handoff to a ground rover offline.
Judging Criteria (How to Win)
  • Mesh Survival: Discovery and swarm formation under constraints.
  • Decentralized Logic: Zero reliance on a single ROS master node.
  • Robustness: Graceful degradation and recovery with latency/dropouts.
  • Developer Clarity: Reproducible simulation environment.
Track 3

The Agent Economy

Building Leaderless Coordination for AI Agents at Machine Speed

Remove the "Master Orchestrator" and let AI agents discover each other, negotiate work, execute it, and prove what happened—all without a central brain.

Build an Agent Coordination Layer using Tashi primitives so that multiple agents can: Discover & Form (temporary swarms), Negotiate & Commit (solve task allocation leaderlessly), Execute & Prove (pass state securely and produce an event record). Minimum demo is 3 agents completing a full loop.

The Tashi Integration Pillars
  • Build agents in JS or Python using FoxMQ (leverages Vertex).
  • For granular control, build in Rust or C using the Vertex library directly.
Example Scenarios
  • Decentralized Agent Marketplace (A2A Bidding): Multiple agents bid and negotiate over complex prompt sub-tasks.
  • Swarm Threat Intelligence: Mesh of cybersecurity agents validating node threats via consensus.
  • Distributed RAG (Hive Memory): Agents simultaneously scraping web sectors, maintaining a synced world view.
Judging Criteria (How to Win)
  • Coordination Correctness: No double assignments. Deterministic resolution.
  • Resilience: Operates gracefully under message delays.
  • Auditability: Complete, verifiable Proof of Coordination.
  • Security Posture: Resistance to replay attacks / message integrity.

The future of autonomy is peer-to-peer

Build the coordination fabric that next-generation robots, drones, AI Agents, and IoT fleets will assume by default.

Register Now → Back to Home

Questions?

Email us at hackathon@tashi.network