Nectar SDK¶

Nectar SDK is a ROS 2 software development kit for autonomous drones. It unifies flight control, computer vision, and AI behind one consistent set of interfaces, so a mission you write once runs across different vehicles, sensors, and simulators with little change.

Get started Architecture

See it in action¶

Autonomous missions our team flew with the SDK, in competition and in the field. Click any clip to watch it larger.

IMAV 2025, indoorGate entry, obstacles, moving smoke platform · 3rd place

SAE Eletroquad 2026Hook and place, annotated run

IMAV 2023, indoorLine following and ArUco landing · 3rd place

CBR 2025Phase 1: Precision landing on a base

SAE Eletroquad 2025Slalom mission

What you can build¶

Fly any vehicle

ArduPilot and PX4 over MAVROS, direct MAVLink, or native uXRCE-DDS, plus Bebop and Crazyflie, behind one flight interface. Navigation, GPS waypoints, return to launch, PID, and obstacle handling.

Fly a drone

See with any camera

USB, RealSense, OAK-D, Pi Camera, or a ROS topic behind one camera factory and a shared image handler. ArUco, color, line, distance estimation, optical flow, and MediaPipe.

See with a camera

Detect and segment

Object detection and instance segmentation across Ultralytics YOLO, HuggingFace DETR, and RF-DETR behind one interface built to take on new models and tasks, with training and evaluation.

Detect and segment

Fly indoors without GPS

A companion rangefinder bridge and a RealSense with Isaac VSLAM pipeline feed the flight controller, so the same navigation code and precision work indoors and out.

Localization

Operate without code

A desktop app to arm, take off, fly, stream cameras with live filters, tune PID with live plots, and inspect ROS 2 topics, services, and parameters.

Interface

Test in simulation

ArduPilot or PX4 SITL with Gazebo, indoor and outdoor, running the same missions you fly on hardware.

Simulation

Who it's for¶

Nectar SDK suits academic labs, research groups, competition teams, and industry prototyping: anyone building autonomous flight who wants control, vision, and AI under one consistent set of interfaces. It is a good fit when you want to:

Run one mission across different vehicles and transports, in simulation or on hardware.
Combine flight control, computer vision, and AI in a single codebase with typed, consistent APIs.
Fly indoors and outdoors with the same navigation code and the same precision.
Add a new drone, camera, or model by following one familiar pattern, without forking the core.

It is not a flight controller or a replacement for ArduPilot or PX4. It runs on the companion computer and drives them through ROS 2.

About¶

Nectar SDK is developed by Black Bee Drones, Latin America's first academic autonomous drone team, founded in 2014 at the Federal University of Itajubá (UNIFEI). The team builds autonomous aircraft for missions that rely on computer vision and artificial intelligence, and competes nationally and internationally, including IMAV (3rd place indoor in 2023 and 2025), the CBR RoboCup Flying Robots League, and SAE Eletroquad.

It started in 2023 as a way to stop rewriting the same camera, PID, and detection code for each competition mission, and grew into a ROS 2 package with consistent interfaces across flight control, computer vision, and AI. It is open source under Apache 2.0 so other teams and labs can build autonomous systems more quickly.