Embedded Vision
for Robotics

From AMRs to humanoid robots

How embedded vision enables real-time perception and action for autonomous systems

Technical progress in robotics and embedded vision is inextricably linked: cameras and real-time image processing form the basis for environmental perception and interpretation. The technology enables autonomous decision-making and learning in the first place. It creates the foundation for self-driving cars and machines, for cobots, and for humanoid robots. Development is advancing at an enormous pace – for the capabilities of the robots as well as for the cameras and embedded vision systems that provide and process the necessary data.
At the same time, these modern robotics applications place the highest demands on the embedded vision components used: spatial perception requires stereo vision; autonomous decisions need real-time computing power. Edge AI or Physical AI enable the execution of multimodal generative artificial intelligence directly on the device. In addition to the corresponding computing power, ultra-compact and energy-efficient cameras and embedded vision systems are in demand, which function reliably even under harsh environmental conditions.

Key Benefits of Embedded Vision in Robotics

High speed and low latency

through fast data transfer and onboard processing

Autonomy & security

autonomous operation with local data processing

Efficient design & easy integration

through fast data transfer and onboard processing

Robust, scalable & cost-efficient

Robust, scalable & cost-efficient

Vision applications for robots: from industry to everyday life

Embedded cameras and vision systems handle key applications in robotics – from use in gripper arms and production lines in the automotive industry to mobile robots for diverse fields of application, as well as cobots, medical robots, and service robots. Here we present some typical embedded vision applications in robotics.

Mobile Robots

Navigation and obstacle detection for AMRs and self-driving machines.

Humanoid Robots

Spatial perception and autonomous action, even in unfamiliar environments.

Industrial robots

Robots in industry handle complex manufacturing processes or work hand-in-hand with humans.

Medical Robots

Robots for use in medical applications assist.

VC Solutions for Embedded Vision in Robotics Systems

From flexibly integratable camera modules and embedded vision systems that integrate image capture and processing, to ready-to-install vision sensors: Choose the right components for your robotics application.

MIPI Camera Modules

Our MIPI camera modules combine maximum flexibility with a wide range of sensors. They can be used with all processors featuring a MIPI CSI-2 interface. We provide the matching drivers and support you in choosing carrier boards and accessories.
board level camera with remote head VCSBC EvoCam

Embedded Vision Systems

Our intelligent all-in-one cameras with integrated CPU and Linux OS are based on modern ARM processors and FPGAs. You have the choice between variants with onboard and remote-head sensors – comprehensively adaptable for your project.
2D/3D laser profiler VC nano 3D-Z - open

3D Camera Systems

The 3D camera systems for
triangulation applications are available for various working ranges from regular to XXL, and upon request, can be customized and provided both with and without housing.

How can we help integrate vision into your robotics project? Let’s talk about your goals and requirements.

  • Selection of the right image sensor
  • Integration: camera module, components, or complete system?
  • Design-in, review, and series qualification
Sven Klette-Matzat - Sales Manager - OEM Solutions - Vision Components

Vision Components: Your partner for Embedded Vision in robotics

VC is a developer and manufacturer – not a distributor. Customers speak directly with experienced Embedded Vision experts and receive the components for their project from a single source. Individual adaptations and variants are possible at any time, perfectly matched to the respective application.

VC Case Studies: Vision Components at work

Cameras and systems from Vision Components have proven themselves in many projects. Here we present some examples – and show why our partners chose VC.

Autonomous mobile robot arculee - Navigation with VC MIPI embedded cameras

Precise navigation for arculus AMRs

Positioning autonomous mobile robots with millimetre precision – using VC MIPI cameras.

Forklift truck drives through a scangate with logistics barcode scanner

Barcode Scanner for Logistics

Prototype with FPGA hardware accelerator VC Power SoM.

Header Schweißroboter sicher führen - Laserschnittkamera

Guiding welding robots: the igm iCAM-PRO laser-cutting camera

Autonomous laser sensor based on a VC board camera

Forklift truck drives through a scangate with logistics barcode scanner

Barcode Scanner for Logistics

Prototype with FPGA hardware accelerator VC Power SoM.

Header Schweißroboter sicher führen - Laserschnittkamera

Safe guidance for welding robots

Autonomous laser sensor based on a VC board camera

More Information on Embedded Vision applications for robotics

Do you have any questions?

Embedded Vision is a key technology for robots: humanoid robots perceive their environment through cameras and can thus react to it in real time; Autonomous Mobile Robots navigate with millimeter precision via built-in cameras; robots for farming applications or service and cleaning robots independently find paths and objects. Modern robots are hardly conceivable anymore without embedded cameras or embedded vision systems like those from German manufacturer Vision Components.

Ultra-compact cameras such as the VC MIPI Camera Modules from Vision Components, which are available with numerous image sensors, are suitable for Autonomous Mobile Robots. They can be easily and flexibly integrated and connected to all common embedded processor boards, for example from NVIDIA or NXP. Important factors for the selection are high quality and robustness, first-class driver support, and long-term availability.

Stereo vision (or stereoscopic vision) is an image processing method that imitates human depth perception. In robotics, the ability to understand the environment three-dimensionally is the key to autonomy and safety. While a single camera (monocular vision) only provides flat 2D images, stereo vision gives robots a true spatial awareness.

 

For stereo vision recordings, two cameras are placed at a defined, horizontal distance from each other. They capture the same scene from slightly different angles. From the minimal differences between the two images—the so-called disparity—the system can precisely calculate the distance to objects in space.

Edge AI is the ability to execute AI tasks such as object recognition, path planning, or speech processing directly on the device—in this case, a robot—without cloud access or external computing units. For this purpose, embedded vision systems such as the VC EvoCam from Vision Components, which enable onboard data processing with AI, are integrated into a robot as hardware. For humanoid robots and other high-performance applications, NVIDIA boards, for example, are used, to which VC MIPI Modules are connected as cameras. This is also referred to as Physical AI when the robots process data from numerous sensors in real time and react directly.

Humanoid robots require cameras and image processing systems that are as powerful, compact, and energy-efficient as possible. These factors are crucial for real-time data processing with AI and long-term mobile use. Furthermore, all components must meet the quality requirements for use in professional series products. The Vision Components cameras and embedded vision systems are therefore developed and produced in Germany. They are also available long-term, so that robots and other series products can be manufactured without time-consuming re-designs.

Physical AI refers to systems that understand the laws of physics, perceive their environment, and can actively interact with it. Humanoid robots, cobots, and self-driving cars are examples of this. They use multimodal AI models to translate the data from numerous sensors into precise physical movements. Cameras and embedded vision systems for onboard data processing play a key role here because they offer a very high depth of data and analyzability. Vision Components is a manufacturer that develops and produces both MIPI cameras, which can be integrated as the "eyes" of the robot, and complete embedded vision systems, which can, for example, be integrated directly on the gripper arm of a robot to handle local tasks there.

Medical robots and surgical robots require ultra-compact camera sensors of the highest quality. Long-term availability also plays a central role here, so that the medical devices can be produced over the long term without re-design and re-certification after the complex development period and certification. Vision Components MIPI Cameras and embedded vision systems, which are specially developed and manufactured in Germany for these requirements in professional series products, are particularly suitable for this. They are available with numerous different image sensors.

Registration

Hands-on Workshop: FPGA Programming for Embedded Vision

23.04.2024

9:30 – 17:00 Uhr

Ettlingen near Karlsruhe

149 € per participation

Don't miss any more VC news

Sign up for our newsletter and be the first to know all our updates!