Research Drone

Research and teaching drone

Specifications

Weight (approx.): 1550g (without battery), 1950g (with battery)

Size: 335mm*335mm*230mm (length*width*height)

Diagonal wheelbase: 410mm

Maximum payload: 1600g

Maximum take-off weight: 3550g

Maximum flight time: Outdoor 13min39s; Indoor 10min23s; Hovering accuracy: ±1m (under GPS) Wind resistance class: 3-4

Working environment: indoor/outdoor

Working temperature: 6℃-40℃

Model: Jetson Xavier NX

Hashrate: 21 TOPS

CPU: Six-core NVIDIA Carmel ARM v8.2 64-bit

GPU: NVIDIA Volta architecture, 384 NVIDIA CUDA cores + 48 Tensor cores

Storage: 64GB

RAM: 8GB DDR4

Display: mini HDMI

Interface: mini PCIe*1; USB 3.0 TypeA*1; USB 2.0 TypeA*1; Micro USB*1; Ethernet port*1 Others: SPI+GPIO multiplexing*1; I2C; UART

Battery：

Model: 4S 4000mah 1piece

Effective capacity: 4000mAh

Battery weight: 374g

Use voltage: 14.8V-16.8V

Minimum voltage: 14.8V

Storage voltage: 15.2V

Battery interface: XT60

Battery type: LIPO lithium battery

Working temperature: 0℃-60℃

Remote controller：

FS-i6S

Binocular camera:

Model: Intel Realsense T265

Chip: Movidius Myraid2

Field of View: Two fisheye lenses combined with a nearly hemispherical 163±5.

IMU: BMI055 Inertial Measurement Unit allows accurate measurement of device rotation and acceleration

Depth Camera:

Model: Intel Realsense D435i

Depth Technology: Active Infrared Stereo

Depth stream output resolution: up to 1280*720

Depth stream output frame rate: up to 90fps

Minimum depth distance: 0.1m

Lidar：

Model: LDS-50C

Laser ranging technology: TOF

Measuring radius: 0.15m~40m

Sampling speed: 9200 times/second

Ranging accuracy: 2~10cm (typical value 5cm)

Ranging resolution: 1cm

Scanning angle: 360°

Scanning frequency: 7~15Hz (typical value 10Hz)

Monocular camera:

Model: USB Monocular Camera

Focal length: 3.6mm

Field of view: 90°

Pixels: 2 million

Maximum resolution: 1920*1080

Software parameters

Board PC：

Model: Jetson Xavier NX

System: Ubuntu 18.04 Username: amov Password: amov

Jetpack: 32.4.3

Cuda: 10.2

ROS: melodic

opencv: 3.3.1

realsense SDK: 2.2.36

realsense ROS: 2.2.15

Prometheus software：

version：v1.0-stable

5：Features

Software part: The prometheus system adopts modular design and is divided into four modules, flight control module, slam module, planning module and target detection module. The P450 product combines the prometheus system to achieve a total of eight indoor and outdoor functions, indoor pointing, indoor landing, indoor tracking, indoor obstacle avoidance, outdoor pointing, outdoor landing, outdoor tracking, outdoor obstacle avoidance, etc. Realize the perfect fit from simulation to real machine.

Achievable functions: indoor and outdoor binocular VIO fixed-point hovering, indoor and outdoor lidar obstacle avoidance, indoor and outdoor vision-guided landing, indoor and outdoor frame detection and tracking, GAZEBO simulation, fastplanner visual obstacle avoidance

P450 function introduction:

Control module

1. Realize the communication between the onboard computer and the flight controller with the help of the MAVROS function package. 2. Provide a simple and convenient control command interface, including one-key unlocking, take-off, hovering, landing,

Emergency locking, control commands (position control, speed control or combined control) under the inertial system and machine system, trajectory tracking, etc.

3. Detailed wiki documentation instructions and code comments, integrated development environment, no need to modify the PX4 firmware source code.

Slam module

Provide rich mapping demos, including RTAB-Map, Octomap and D435I+T265 environment mapping

Planning module

1. Provide a wealth of planning demos, including APF local planning, VFH local planning, and Astar global planning

2. Fastplanner implementation and trajectory optimization demo

Target recognition module

1. Image recognition based on Opencv, providing rich demos, including digital recognition, ellipse

Detection, multi-QR code detection, color recognition, YOLO general target detection, frame selection tracking, etc.

2. Detailed wiki documentation instructions and code comments.

Task module

1. Combine the above modules to provide simple and rich application-layer application cases, including autonomous landing, circular frame penetration, waypoint tracking, color line tracking, digital recognition, target frame selection tracking, waypoint planning and other demos.

2. Provide detailed wiki instructions and code comments, easy to modify and highly reusable.

Gazebo emulation

1. The simulation environment software system can use virtual engine technology to provide a realistic environment, simulated power and sensing, and a built-in flight controller, which can simplify the initial setup process and make UAV simulated flight easier. This function can improve test test efficiency through control and state estimation algorithms, which is more advantageous than embedded high-cost debugging and development.

2. P450 provides simulation tutorials corresponding to real machine functions, with detailed wiki instructions for use

Various problems encountered in the simulation environment.

Hardware part: Modular design, which can be flexibly matched with airborne equipment. The composite material body can effectively reduce the weight of the whole machine and effectively prolong the battery life. The semi-closed fuselage design further improves the protection ability and effectively reduces the interference of airflow