This ArduPilot Developers Conference 2025 recap features a series of presentations and live demonstrations showcasing the latest advancements and community contributions. Key updates include Andrew Tridgell's overview of new ArduPilot features like advanced CAN and network protocol capture, a web-based DroneCAN GUI, a comprehensive overhaul of the support proxy for secure, multi-client connections, and a deep dive into LTE modem integration for companion-computer-less BVLOS operations, inspired by an autonomous sailing buoy.
Further highlights include Eric Maglo's impressive F-35 VTOL demonstration, a detailed update on Gazebo simulation capabilities and new models for various vehicle types by Reese Mainwearing, and Andreas's review of Mission Planner updates, introducing new gimbal controls, advanced terrain visualization, and a sneak peek into upcoming European no-fly zone handling. The conference also presented a novel FFTD gyro system for indoor drone simulation and tuning, and a captivating plane follow demonstration using ArduPilot's robust follow library. The day concluded with a high-stakes "Doritocopter" demo, showcasing a 32-motor aircraft's resilience to multiple in-flight motor failures, emphasizing ArduPilot's redundancy capabilities. Throughout, the emphasis was on community collaboration, open-source development, and pushing the boundaries of autonomous flight.
Exploring the ArduPilot Developers Conference 2025: Innovations in Autonomy and Simulation
The ArduPilot Developers Conference 2025, hosted by 3DXR in North Yorkshire, is introduced, along with sponsors and essential logistical information for attendees, including health and safety, facilities, and the schedule.
Details for flight activities are provided, including flyer ID requirements, area familiarization, weather outlook, and airspace restrictions, with fun flying planned for the evening.
0:34:10 Andrew Tridgell - ArduPilot Update: New Features & Demos:
0:35:18 Community Growth: Tridgell highlights an active year for ArduPilot, with over 5,500 patches, 2,600 pull requests, 265 contributors, and 77 new board ports, totaling 392 hardware definitions.
0:37:03 Advanced Protocol Capture: Introduces generic CAN protocol capture, logging all CAN packets to bin logs for driver development and longevity testing of CAN servos through simulation.
0:40:14 Network Packet Capture: Details built-in packet capture for Ethernet and PPP connections, generating Wireshark pcap files directly from the flight controller for debugging network-enabled devices.
0:43:56 Web-based DroneCAN GUI: Announces a new web-based GUI for DroneCAN by Hoy Bean (Bim Drones) for easier accessibility, with a live demo attempting connection via an encrypted web socket over an LTE modem.
0:45:52 Support Proxy Overhaul: Presents a major redesign of the support proxy, now supporting multiple client connections (UDP, TCP, WebSocket, WSS with Mavlink signing) to a single vehicle stream, improving remote operation efficiency.
0:50:41 Ambitious Live Demo & LTE Integration: Tridgell attempts to connect over 100 audience members to a rover via the support proxy, simultaneously explaining the integration of low-cost LTE modems for companion-computer-less BVLOS operations, inspired by his autonomous sailing buoy project. The demo faces connection issues due to high demand.
0:53:00 JavaScript Mavlink Bindings: Showcases new bidirectional JavaScript Mavlink bindings with Mavlink 2 signing and SSL support, enabling secure custom web UIs and ground stations.
1:09:12 Multiple PPP & Routing: Highlights support for up to three PPP links and IPv4 routing capabilities within ArduPilot, allowing complex network configurations.
2:08:32 F-35 VTOL Flight: Eric Maglo demonstrates his custom ArduPilot quadplane F-35 VTOL, featuring a unique motor mixer for vertical takeoff, seamless transition to forward flight, and a smooth vertical landing.
2:22:48 Gazebo Overview & New Features: Reese provides an update on Gazebo, an open-source robotics simulator, highlighting new Python bindings, advanced physics (e.g., for marine vehicles), new sensor types (air pressure, Doppler logs), and more realistic camera models.
2:31:00 ArduPilot Plugin Updates: Mentions updates to ArduPilot's Gazebo plugins, including camera zoom, GStreamer streaming, gimbals, and corrections for simulation reference frames.
2:36:51 Expanding Model Library: Showcases new ArduPilot models in Gazebo, including Mini Talon planes, AION R1 rovers, unique Omni Rovers, Blue Boat marine vehicles, a Land Yacht, and models for bicopters and slung payloads.
2:49:15 DroneCAN for Simulation: Emphasizes using DroneCAN to pipe sensor data from Gazebo into ArduPilot, making simulated sensors appear identical to real ones.
2:51:38 Helicopter Simulation Breakthrough: Demonstrates a surprisingly plausible helicopter model in Gazebo using rigid body physics, even achieving auto-rotation capabilities after tuning specific joint constraints and adding virtual sensors.
3:08:14 Mission Planner Status & Development: Andreas confirms Mission Planner's ongoing development with numerous updates and contributors, advising users to utilize beta releases for the latest features.
3:09:49 New Gimbal Interface: Introduces a new gimbal interface by Robert Long, allowing video feeds to be integrated into the map or a separate window, with direct gimbal control and customizable shortcuts.
3:11:58 Advanced Terrain Tools (Ctrl+W): Unveils enhancements to Mission Planner's terrain tools, including dynamic terrain visualization with reference scales, elevation views relative to the vehicle, and line-of-sight propagation for radio.
3:15:51 Custom Terrain Data & Maker: Highlights the ability to integrate custom digital elevation models (GeoTIFF) and a new integrated terrain maker for generating high-resolution terrain data files directly within Mission Planner.
3:20:41 Waypoint Altitude Warning: Implements a critical warning for waypoints set to zero altitude, preventing common user errors that could lead to crashes.
3:22:13 Third-Party Plugins: Showcases powerful plugins from Yuri Rage (Planner Macro, Fence Replanner, Optimal Coverage) and Andreas's own (Messages NG for enhanced logging, Mission Planner Starter for multi-configuration management).
3:40:11 Upcoming: European No-Fly Zones: Previews upcoming support for European no-fly zone data based on a JSON schema, crucial for drone certification in Europe, including dynamic restrictions, proximity warnings, and geo-awareness status.
4:48:21 Autonomous Plane Follow: Tim demonstrates a second plane (T2 Cruiser) autonomously following a lead plane using ArduPilot's advanced "follow" library and a Lua script, showcasing complex autonomous coordination and real-time adjustments for wind.
4:51:45 Future Development: Tim discusses ongoing refinements to the Lua script and plans for future features like VTOL auto takeoff/follow/land.
5:02:59 Simon Hancock - Linking Drones Presentation:
5:04:58 Origin & Unique Design: Simon shares the story of Linking Drones, detailing Alberto's novel concept of a VTOL tail-sitter drone with a payload chassis acting as a wing, controlled by only four asymmetric rotors.
5:06:33 Market Shift & Prototype Validation: The company shifted from consumer to business focus, and early RC-flown prototypes validated the unconventional design, highlighting ArduPilot's flexibility.
5:10:31 User-Friendly Control & Architecture: Emphasizes control via a mobile app using Bluetooth for simplified map-based automissions, with an RC radio as backup. Simon outlines the drone's architecture, including its onboard mini-computer and bespoke messaging protocol.
5:17:40 Development Challenges & Field Testing: Describes challenges like managing conflicting arm/disarm commands and terrain data. He recounts a trip to the Philippines for remote medical delivery trials, facing numerous field testing adversities.
5:23:24 Center of Gravity & Load: Addresses the drone's center of gravity with payload and its load capacity (1.8kg empty, max 1kg payload).
5:40:50 EKF Basics & Motivation: Rishab clarifies how ArduPilot runs multiple EKFs ("lanes") for each IMU and explains EK3_SOURCE parameters, detailing the motivation for new work stemming from issues with a new computer vision sensor.
5:47:17 New "Source per Lane" Feature: Introduces EK3_SOURCE_OPTIONS (bit 3), allowing each EKF lane to use a different EK3_SOURCE parameter set, enabling concurrent testing of multiple sensor configurations.
5:49:33 Lane Switching Enhancements: Explains existing automatic lane switching and introduces EK3_OPTIONS (bit 2) for "Manual Lane Switching," giving explicit control over primary lane selection via Lua scripts.
5:52:21 General EKF Fixes: Details various EKF improvements, including correct vertical velocity fusion for external nav, accurate variance usage, and a new feature to revert to the primary compass after landing.
5:57:16 Future Work: Outlines plans for better Lua access to EKF core health, a rework of existing lane switching logic, the ability to fuse multiple position sources, and separating EKF cores from individual IMUs for greater flexibility.
6:44:56 World's First 32-Motor Copter: The conference concludes with a demonstration of a "Doritocopter," the first known 32-motor aircraft to fly with ArduPilot, built to showcase redundancy.
6:45:57 Motor Failure Test: The aircraft, with 32 PWM motors split over CAN, performs an auto mission while a Lua script randomly switches off motors, demonstrating its ability to maintain flight despite significant in-flight motor failures.
6:51:07 Successful Recovery: The Doritocopter successfully recovers, changes flight modes, and lands after multiple motor failures, concluding the planned demos.