Computer vision provides fundamental features for smart robots, such as object detection, tracking, navigation and mapping (SLAM), obstacle avoidance or grasp planning. We offer consulting, design and implementation for your computer vision project, both with "classic" algorithms and deep learning. Our expertise derives from 15 years of research at Visevi and the Technical University of Munich.
We design and implement demonstrators and prototypes of smart robotic systems for lab use cases, innovation studies and pilots production. Our prototypes are commonly shown on trade fairs to evaluate customer interest. Besides software development, our work also includes mechanical design, rapid prototyping and integration.
ROS (Robot Operating System) is the way to go for smart robots. Cutting-edge algorithms and technologies are readily available, open source. With ROS2 and ROS-Industrial, the codebase is ready for industrial use. We have been working with ROS for 10 years and have published and contributed to numerous packages. We assist you with consulting, system design, development and component integration.
Perception and scene understanding are fundamental steps for any smart robot. We are experts in state-of-the art computer vision algorithms and in making them work in real-time on embedded systems, GPUs or standard computers.
Depth and RGBD sensors are now widely available, also thanks to low-cost products like the Realsense. We support you with processing and interpreting depth data using cutting edge algorithms for applications like navigation, trajectory planning and grasping.
The performance of algorithms in computer vision, grasping, planning and many more has been revolutionized in recent years with deep learning and other AI technologies. Bring your system to the next level with Visevi's support!
Grasping and picking of different or sensitive objects is a real challenge for robots. We consult you in finding the right algorithms, grippers and vision systems, and implement customized solutions using our bin picking, grasp planning and tactile sensing software packages.
Modern user interfaces for robots or other smart devices are web-based – and thus usable on any smartphone or laptop via WiFi. We integrate real-time data from the device, even including video and 3D, using modern HTML5 technologies. With our Webcontroller software, we prototype your UI fast with a low budget.
The Robot Operating System (ROS) offers great connectivity and endless features for robots. Whether you are just getting started or have already a complex setup – we design and integrate your system professionally.
With our expertise both in low-level HW/SW interfaces and in ROS, we get any component connected with an adapted and fast ROS driver/interface. Check out our Github site for some examples.
We typically work on projects according to the following roadmap, which we discuss and adapt with you as needed. This process allows to show results early-on and to reduce project/budget uncertainties, which is especially important when working with advanced technologies. At each stage, you get valuable details and you decide how the project shall continue. Click on the stages to learn more.
In an initial feasibility study, we identify and analyze available products and technologies for your project. You will know what's possible for what budget early-on, even when working with technologies from research.
We implement a prototype or PoC of critical system components that can already be presented as a running demonstrator. This helps to convince colleagues and customers, and also serves as a valuable basis for "hands-on" experiments to define the further project.
We implement your project using the agile development approach with continuous integration. This assures high code quality, quick adaptation to refined requirements and targeted development of the features you really need. Your team can also participate the development and always stays in the loop.
Count on our support all the way: We perform the final integration of the project, making sure it work with your systems. Of course, we provide detailed documentation and coach your team on how to use the project results most effectively.
Intelligent bin picking software with AI-based grasp planning for sensitive and soft objects. Demo with Keba, using ROS and ViseGrip.
ViseGrip in a pick-and-place application developed together with Gimatic: Various objects made of glass and plastic are picked from the white holder, detected an dsorted based on camera-based tactile sensing