Blätter-Navigation

Offre 207 sur 267 du 19/07/2022, 16:56

logo

Bark­hau­sen Insti­tut gGmbH - Inter­net of Things

The Barkhausen Insti­tute per­forms research on the latest hard­ware, soft­ware and com­mu­nic­a­tion tech­no­lo­gies and demon­strates their applic­a­tion in tomor­row's soci­ety. Our goal is to build a flex­ible and trust­worthy IoT plat­form based on reli­able elec­tron­ics and secure com­mu­nic­a­tion meth­ods.

Stu­dent Assist­ant/Pro­ject work (m/f/d)

Machine learn­ing based con­trol sys­tem for line fol­low­ing cars

Working field:

The pro­ject

The cars on our roads have become more and more smart and equipped with sensors. But one key ingredi­ent which is still miss­ing is the com­mu­nic­a­tion between the cars. To show­case the import­ance of this aspect the Barkhausen Insti­tut has developed a demon­strator show­ing the effects which proper com­mu­nic­a­tion between vehicles can have on the traffic on the road.
This demon­strator con­sists of RC vehicles driv­ing along a pre­defined track. They keep their tra­ject­ory by fol­low­ing a line on the track. Cur­rently this line fol­low­ing is done using a cam­era moun­ted on the vehicles and a PID based con­trol loop based on com­puter vis­ion. Your task is to replace this com­puter vis­ion applic­a­tion with a machine learn­ing approach. The Don­key­Car¹ Frame­work is espe­cially developed for such tasks. Your task is to imple­ment a new con­trol sys­tem using the Don­key­Car Frame­work and do a object­ive com­par­ison of the per­form­ance of both approaches, regard­ing track­ing sta­bil­ity, sturdi­ness against envir­on­mental factors, such as dif­fer­ent light con­di­tions, and highest pos­sible driv­ing speed.

Your tasks
  • Learn­ing about the basics of machine learn­ing in the con­text of com­puter vis­ion
  • Imple­ment­ing a cus­tom ML-based con­trol sys­tem in the exist­ing demon­strator
  • Com­par­ison of the old approach to the new ML approach

¹https://www.donkeycar.com/

Requirements:

  • Study of Elec­trical Engin­eer­ing or related courses
  • Good com­mu­nic­a­tion and writ­ing skills in Eng­lish
  • Work­ing with Linux and the com­mand line envir­on­ment
  • Basic under­stand­ing of con­trol loops and their per­form­ance meas­ures

What we offer:

  • An open and cre­at­ive team of enthu­si­astic sci­ent­ists
  • Close to cam­pus
  • Agile work­ing hours and work­flows
  • Work on future tech­no­lo­gies

If you have pas­sion for new tech­no­lo­gies and the desire to develop the con­nec­ted world of tomor­row with us, we are look­ing for you.

Life and sci­ence are com­pat­ible at our insti­tute. We offer a range of recog­nised fam­ily-friendly meas­ures, such as part-time work, mobile work­ing and flex­ible work­ing hours that take fam­ily needs into account. We also offer sup­port for arriv­ing in Dresden (all form­al­it­ies, apart­ment etc.).

How to apply:

Short let­ter of motiv­a­tion and cur­riculum vitae to careers@barkhauseninstitut.org. We only accept com­plete doc­u­ments as one single PDF doc­u­ment by e-mail. Please ref­er­ence “Stu­dent Assist­ant-CoRoLa” in the sub­ject line. Applic­a­tions that con­tain other file formats (e.g. Word, Excel) will not be con­sidered in the selec­tion pro­cess. Informal inquir­ies can also be sub­mit­ted to the same address.

The data sub­mit­ted by you in the course of the applic­a­tion pro­ced­ure will be pro­cessed by the Barkhausen Insti­tut (BI). BI shall use the data for the pur­poses of the applic­a­tion pro­ced­ure. No data shall be passed on to third parties.