Internships at PAL Robotics: Bence Magyar reports

Bence Magyar Internship Report

Hello there!

I’m Bence Magyar from Eötvös Loránd University from Hungary. I’ve spent the last 5 months working at PAL Robotics as an Erasmus intern in Computer Vision.

Bence Magyar, REEM and PAL Robotics’ employee Jordi Pages (PhD in Computer Vision)

The primary goal of my project was to help REEM, the Humanoid Service Robot from PAL Robotics, grasp objects. To be a bit more specific: I had to develop a system which can detect and estimate the location and pose of an object. Since Computer Vision is a really wide and rich field, I had to carefully investigate which path to take to solve this problem. At the beginning of my work I completed a survey of the available techniques where I classified them and did more research and experiments on the most promising candidates. I ended up integrating and improving a work of Vienna University of Technology, namely BLORT.
As a side effect I also produced other components which can help with several different Computer Vision tasks.

Luckily there was some time left at the end of my internship so I had the opportunity to also work a bit on the problem of tracking the hand. Tracking the hand is crucial for visual servoing applications such as the one Joven implemented.

Bence Magyar’ self portrait

As a software person I’d like to give credit to the open-source software I used during my internship at PAL Robotics. The “engine” of the robot is ROS, the work I based my project on is BLORT, my operating system of choice is Ubuntu

My work is available here:
Perception Blort
PAL Vision Segmentation

I will also prepare a video of my project soon, stay tuned!

Living in Barcelona and working at PAL Robotics was a great experience!
I’m going back to Hungary for the next semester to finish my studies in Computer Science and to write my thesis manuscript on the work I did at PAL.
Thank you guys for these great months!

Cheers,
Bence

Related articles
TIAGo Pro, the mobile manipulator designed by PAL Robotics in the food industry
Read More

ROSALYA: advancing mobile manipulation and smart automation in the food industry

The food-processing industry faces one of the greatest challenges in modern manufacturing: ensuring high productivity and quality in environments where…
StockBot in Picking Warehouse
Read More

Warehouse inventory accuracy with StockBot: Smarter picking, fewer errors

In today’s picking warehouses, where thousands of items move in and out every hour, the success of an operation depends…
How RoboSapiens is building trustworthy self-adaptive robots for industry 5.0
Read More

How RoboSapiens is building trustworthy self-adaptive robots for industry 5.0

As robots take on more complex roles in industries, cities and people’s daily lives, one of the biggest challenges  has…