We at kleinlabs would love to help you and your customers get rid of dull, boring and error-prone tasks. While the news catches a lot of attention with the most challenging automation — e.g., autonomous driving — there are many applications where simpler automation can tremendously help. Think for example about how barcodes help reduce cashier load.
Do you have a task that requires humans to mechanically look at something and type data into a database? Maybe to count the remaining number of items in a box? Or count the number of cars in a parking place?
We are here to help. Get in touch with us to discuss how to automate such tasks.
How to work with us?
We can adapt our collaboration to your needs, but here is a workflow that is a good start:
- Send us videos describing the task you want to automate. The more the better. We prefer two types of videos:
- “Far” videos allowing us to understand the problem, as if taken by a movie director.
- “Near” videos as seen by the human performing the task or the future automation system. The following characteristics are desirable for near videos:
- Fixed position(s), fixed zoom
- In focus
- Well-light environment
- Various conditions (e.g., day, night, early morning)
- Various scenarios (e.g., box is inserted in container, box is removed from container)
- We will evaluate under strict NDA:
- What is the potential for automating the task?
- What computer vision technologies and methods can be used?
- What hardware is most cost-effective at solving the problem (e.g., lenses, sensors, embedded computer vs. powerful server)?
- We then sit down together and draft a project.
- If the task is easy to automate, we can already specify the system and implement it for you.
- If the task is difficult to automate, we can suggest:
- More hardware configurations that can be tested.
- More software techniques that can be tested.
- Human-in-the-loop automation to do 99% of the automation, while letting a human do the difficult 1%.
Unsure what can be automated and how the final system should look like? We are happy to help you specify the project. We like challenges and will never give up automating a task, but at the same time will ensure that you are in control of the costs.
We do more than algorithms!
Although the first throw is about the algorithms and the computing platform they are running on, we are able to do more:
- Housing: We are able to design your prototypes as we did for the FreezFiler solution and are even able to 3D print the prototypes in small volumes. This let’s us redefine together your needs and application and provides an incredible short round trip time for new iterations on the planned product.
- PCB Design: When we need small add on to the existing single board computers, we are able to design suiting PCB (printed circuit boards) in house. Once again your advantage to fast solution at a single shop.
We are near you!
We are based in Umeå/Sweden and in Berlin/Germany. Both cities provide excellent connection to whole Europe and we are able to get connected to you by the communication means of your choice
Is it expensive? Not with us!
We know that a lot of interesting applications start to get unrealistic when a first calculation of components lead to a price tag that is not in relation with the problem to be solved. Therefore we decided to go strategically with off-the-shelf components that are economic, easy to replace with compatible components (second source suppliers) and still reliable.
Using the computing performance of single board systems which are state of the art, we combine affordable hardware components with our software and system-design know-how to provide you with efficient, economic and reliable solutions.
The following list should give you an idea of components we are comfortable with. If you already have an idea of other supplies, we could review if we are able to port our modules to them:
- Sony IMX219
- OmniVision OV5647
- ELP USB cameras
- Server-class hardware
- Raspberry PI
- Tinker Board (RK 3288)
- Firefly boards (e.g., RK 3288 or RK 3299)
- Rock 64 (RK3328)
- Linux (especially embedded versions, like Rockchip Linux)
- Google Test
- ReactJS, Redux, Bootstrap, Material, Font Awesome