Xamla Logo

(ROSVITA™ + egomo™) = adaptive robotics

Xamla is an internal Start-Up of PROVISIO GmbH, Germany, focussing on adaptive robotics using AI processes. Our goal is - together with the worldwide robotics community - to ring in the era of Personal-Robots. We teach robots to see, feel, and understand the composition of a scene, and enable the programming of intelligent robot behavior.

The open source robot programming IDE "ROSVITA" - in combination with the egomo sensor head - allows you to enter the physical world of adaptive robotics fast and low-cost by minimizing common hurdles.

Hardware Kit Icon

Open Source

At Xamla we believe that only open-source development - not monopolized stand-alone solutions - can help avoid redundant work, and develop cost-effective robot solutions. This is why the Xamla programming and robot control solutions are being developed based on open source and can be used FREE OF CHARGE, even for commercial applications.


The robotic system defined by us can be used for a wide array of tasks. Users have access to an ever evolving and expanding library of applications - no costly and time-consuming individual programming necessary.

Hardware Kit Icon
Hardware Kit Icon

Adaptive sensor head

With its stereo cameras and RGB-D sensors our low-cost sensor head „egomo“ scans the robot's surroundings and creates a 3D model of the real world inside the computer. Objects and obstacles are tracked in real-time. Tasks can be solved situational without the need for the objects that are to be manipulated to be at a specific fixed position.

Hardware Kit

Xamla applications are offered for standardized systems. In its first iteration the egomo sensor head was developed specifically to work with a Universal Robots UR5 in combination with a Robotiq 2-finger-85 gripper, and optionally the Robotiq FT-300 force-torque sensor.

Hardware Kit Icon
Hardware Kit Icon


Xamla in Münster (Westf.), Germany is looking for

Web Developer (German-speaking)

Entwickle bei uns mit an der Browseroberfläche einer größeren Web-Anwendung, mit der sich eine Vielzahl von Aufgabenstellungen aus dem Bereich Computer Vision & Machine Learning vollständig im Browser abbilden und lösen lassen.

All open positions / Details


New video: Duplo Stacking 1.2

We have just uploaded a video of the next iteration of our "duplo stacking" demo (see previous blog entry) which now allows for both the actual duplo bricks as well as the target plate to be located at random locations.

Video: Duplo stacking with egomo

This vision guided pick & place demo uses Xamla's open-source sensor head egomo to pick up two freely positioned Duplo bricks and stack them at a predefined position on a Duplo plate.

The pose of the Duplo bricks is first coarsely measured using a point cloud obtrained from a Structure IO depth sensor and subsequently refined using high-resolution stereo triangulation.

Presentation: Deep learning with artificial neural networks

Like most university towns Münster has a thriving hacker and maker community. We love interacting with these people, and even recruit some of our team members from them. For a presentation on "Deep learning with artificial neural networks" at the local hackerspace "Warpzone" our CTO Andreas Köpf prepared a slideshow which we happily share with everybody interessted in this topic.

From Siri and Skype-Translator to facial recognition and predicting behavioral patterns – deep-learning has become part of our livess. Since 2006 we are in in the third advent of artificial neuronal networks. Thanks to deep and recurring models with millions of parameters, computers are able to decipher handwriting identify objects on images or videos, process natural speech, and much more. Many tasks previously reserved for humans can now be automated.

We'll start with an introduction into how deep neural networks work, and then take a closer look at some of the spectacular results presented in the last couple of months - Is the hype justified? The presentation is aimed at everyone interessted in artificial intelligence; knowledge of neural networks is not required.

Download Presentation (PDF - German)

Older blog entries