CN EN

INDEMIND M1

A hardware of multi-sensor fusion architecture and microsecond time synchronization mechanism specially provided for developers

Product OverviewParametersFAQSDKBUY NOW

INDEMIND M1

INDEMIND M1, a hardware for developers, adopts a "binocular camera + IMU" multi-sensor fusion architecture and microsecond time synchronization mechanism, to provide accurate and stable data sources for visual SLAM research, to meet the development requirement such as SLAM research, navigation and obstacle avoidance development, visual motion capture development, stereo vision development and others.

INDEMIND M1

200FPS high-speed data acquisition

INDEMIND Binocular Inertial Module is the first binocular inertial camera that supports 200FPS image acquisition. It can accurately obtain the image and IMU data under high-speed maneuverability, accommodate high-speed acquisition demands of developers, greatly decrease the tracking difficulty of the related algorithm, and accommodate positioning and mapping data demands of the high-speed SLAM algorithm (e.g. vehicle-mounted).

Binocular camera +IMU

With the camera +IMU multi-sensor fusion structure, the module can realize compensation between the camera and IMU sensor and provide a solution for binocular stereoscopic visual hardware with higher attitude accuracy, stronger environmental suitability, more stable dynamic performance, and lower costs.

High-resolution camera

The high-definition camera with a global shutter can provide 1280×800@25/50/100FPS & 640×400@25/50/100/200FPS image data in the range of horizontal 120° and vertical 75° level, and effectively reduce image distortion detected under motion image to provide strong front-end data acquisition capacity for SLAM algorithm.

Built-in 6-axis IMU sensor

The built-in 6-axis IMU sensor with max. operation frequency 1000Hz can accurately measure motion characteristics of the carrier, refresh attitude information of the carrier in high frequency, and fast create a 3D perception for surroundings.

Binocular +IMU microsecond-level synchronization

The unique high-precision time synchronization mechanism can improve the synchronization accuracy between the binocular camera and IMU to a microsecond level and reduce sensor error to further provide an accurate and stable data source for the SLAM algorithm.

Multi-sensor calibration

The professional calibration is made for multiple sensors such as camera and IMU, of which accuracy is superior to other open-source calibration methods (e.g. Kalibr) to reduce multi-sensor assembly error and further improve the accuracy of data acquisition and system stability.

Kalibr：

组别 head(°) pitch(°) roll(°)
1 -179.5008 -0.6017 0.6507
2 -179.3397 -0.6017 0.6507
3 -179.1031 -0.1903 0.4534
STD 0.2000 0.2211 0.1502

INDEMIND：

组别 head(°) pitch(°) roll(°)
1 -179.521225 -0.439943 0.622170
2 -179.602241 -0.463838 0.598827
3 -179.780491 -0.662679 0.404572
STD 0.1326 0.1223 0.1195
※The experimental results show that the repeatability accuracy of INDEMIND calibration installation angle error is 1.5 times that of Kalibr algorithm, and the camera re projection error is also better than that of Kalibr algorithm.

SDK

Data acquisition
The binocular +IMU structure and USB3.0 transmission characteristics can realize high-speed and stable data acquisition.
Object detection
Based on the deep learning technology, the product can realize semantic, size, and attitude information recognition and detection for environmental objects.
Depth map
The collected depth map can be used for obtaining depth information of objects. Besides, the open-source algorithm can be used for obtaining 3D data of objects.
Point cloud
The dense point cloud map can be created to generate the 3D model of objects.

Hardware technical indicators

  • Applicable distance：

    0.1m-10m

  • Lens：

    Replacable Standard M12

  • Visual angle:

    D：140° H：120° V：75°

  • Resolution:

    1280×800 | 640×400

  • FPS (frames per second):

    25FPS | 50FPS | 100FPS | 200FPS

  • Baseline:

    120mm

  • IMU frequency:

    1000Hz

Developer Support

  • Visual frequency:

    25Hz

  • IMU frequency:

    200Hz

  • Applicable OS:

    Windows、Linux、ROS

  • Calibration file:

    IMU calibration, camera calibration, IMU/camera’s external parameters calibration

  • Developer support:

    SDK

Field of application

Visual SLAM research

Visual SLAM research

The binocular visual-inertial module uses the global shutter camera and 6-axis IMU to realize time synchronization between camera and IMU and reach the microsecond level for accuracy of time synchronization. Besides, the module provides IMU calibration parameters, camera calibration parameters, and external parameters of camera and IMU for high-accuracy SLAM development.

Visual motion capture and development

Visual motion capture and development

Based on the module image data, the product uses the human kinetics algorithm to realize 3D stable and high-precision multi-user skeleton tracking.

Navigation and obstacle avoidance development

Navigation and obstacle avoidance development

The binocular visual-inertial module can be compatible with multiple open-source projects and can provide various map modes (e.g. dense map, point cloud map) to lay an essential foundation for navigation and obstacle avoidance development. It can be widely used for developing wearable devices, UAVs, robots, and AGVs.

Stereoscopic vision development

Stereoscopic vision development

The Binocular Visual Inertial Module adapts to a number of open-source algorithms and enables ultra-high precision recognition and detection over a fixed range, thereby providing users with high-accuracy 3D point coordinates for high-precision scene reconstruction.

Product name: Binocular visual-inertial module

Product specification

  • Product

    Product name: Binocular visual-inertial module

  • Size

    PCBA plate: 95mmx25mm; Overall length and width: 140mmx25mm

  • FPS (frames per second)

    【25FPS】【50FPS】【100FPS】【200FPS】

  • Resolution

    【1280×800】【640×400】

  • Depth resolution

    Based on CPU Up to 640×400@25FPSz

  • Pixel size

    3.0 x 3.0μm(1280×800)

  • Shutter speed

    5ms

  • Baseline

    120.0 mm

  • Lens

    Replacable Standard M12

  • Visual angle

    D:140° H:120° V:75°

  • Focal length

    2.09mm

  • Filter

    N/A

  • Motion perception

    6 Axis IMU

  • Color pattern

    Monochrome

  • Scanning mode

    Global shutter

  • Power consumption

    1.05W @ 5V DC from USB

  • Synchronization precision

    Microsecond level

  • IMU frequency

    1000Hz

  • Output data format

    Raw data

  • Interface

    USB 3.0

  • Weight

    40g

  • UVC MODE

    YES

  • Applicable distance

    0.1-10m

Software

  • Applicable OS:

    Windows10、Linux、ROS

  • Developer support:

    SDK development and document

  • Calibration file:

    IMU calibration, camera calibration, IMU/camera’s external parameters calibration

  • Hardware requirements:

    CPU、USB3.0

Environment

  • Operating temperature:

    -10°C~60°C

  • Storage temperature:

    -40°C~120°C

  • Humidity:

    0% to 90% NON-CONDENSING

Packaging

Packaging:

Packaging: 1× binocular visual-inertial module; 1× Micro USB 3.0 USB cable; User Manual; portable bag

Warranty

Warranty

If the binocular visual-inertial module occurs any hardware fault due to non-deliberate damage within one year as of the date of purchase, after detection by the after-sales service center, INDEMIND will provide free warranty service.

FAQ

E-mail：support@indemind.cn

Which kind of task and scenario are applicable to the binocular visual-inertial module?

With computer vision technology, the INDEMIND binocular visual-inertial module can provide a stable and reliable hardware foundation. Besides, the module can use the computer vision algorithm to accommodate SLAM, deep calculation, object recognition, gesture recognition, face recognition, dense mapping, and obstacle detection under the field of SLAM research, robot, UAM, AR/VR, and AGV.

Will the module support the open-source SLAM project?

INDEMIND binocular visual-inertial module data support common open-source SLAM projects (e.g. ORB SLAM, VINS, DSO, SVO) and support users to create the test set similar to the EuRoc format.

Which kind of data output will SDK support?

Through applicable client software or SDK, the module can support binocular image, IMU data output and storage.

What is the effect of object recognition?

It will recognize an object and measure the position information of the object under the world coordinate system and obtain the 3D semantic map.

Will the module SDK be used for a developing environment?

SDK supports Windows & Linux system environments.

Will the camera realize hardware synchronization with the IMU sensor?

Yes. The built-in high-precision chip can control time synchronization between the binocular camera and IMU sensor to realize accuracy up to a microsecond level.

Will INDEMIND binocular visual-inertial module be calibrated before delivery?

Yes. Before delivery of INDEMIND binocular visual-inertial module, we will make professional calibration for external parameters of camera-IMU (including installation angle, lever arm), camera, and IMU to ensure availability of module.

SDK instructions

View details >

Download SDK

View details >

采集软件

View details >

Instructions for Windows operation

View details >

Instructions for Linux operation

View details >

Instructions for ROS operation

View details >
Vi-SLAM 深度演示

INDMEIND双目视觉惯性模组支持运行多种开源算法，实现高精度定位效果（自研VSLAM暂不对外）

完整视频

INDEMIND双目视觉惯性模组支持输出高精度深度图像

完整视频

开源算法示例

ORB运行演示

DSO运行演示（室内环境）

DSO运行演示（室外环境）

VINS运行演示

技术支持

任何产品相关的技术问题均可和在线客服进行交流，我们将尽快为您提供帮助

  • QQ customer service

    3457489099

  • WeChat Assistant

    WeChat search：INDEMIND

  • Technical service E-mail

    support@indemind.cn