Ontvang aanbevelingen op basis van je profiel en word benaderd door bedrijven.

Graduation: Parallelizing algorithm for autonomous vehicle coordination

Geplaatst 23 jan. 2023
Delen:
Werkervaring
0 tot 1 jaar
Full-time / part-time
Full-time
Soort opleiding
Taalvereiste
Engels (Vloeiend)

Ontvang aanbevelingen op basis van je profiel en vind de vacatures die bij je passen.

Al 300.000+ studenten en starters hebben een profiel.

Start date: Any time

Assignment duration: 6 months

Context

Autonomous vehicles (AVs) bring flexibility to warehousing, airport baggage handling, and parcel sorting. However, to combine this flexibility with high throughput in tight spaces, fine-grained coordination between the AVs is necessary. To that end, Vanderlande has developed a Fleet Management System (FMS). This FMS is responsible for controlling AVs in such a way that the system reaches the highest throughput feasible without collisions or deadlocks.

Problem description

One of our algorithms is a multi-agent path planner based on Cooperative A*; it plans paths of all the AVs fully in space and time for active jobs in real-time. This fully prevents deadlocks and offers possibilities for high optimization. However, due to the algorithm design, this can only be done single-threaded. As this significantly limits the scalability or resulting throughput, we want to extend the algorithm’s capabilities through parallelization.

Tasks / responsibilities

  • Understand the challenges that the algorithm tries to solve
  • Investigate existing state-of-the-art solutions in this domain
  • Design a parallelized algorithm based on the principle of fully planning paths of AVs in space and time
  • Prototype the algorithm in a programming language of choice (whether GPU-based, multi-core, multi-process)
  • Integrate the algorithm into the FMS, including its emulation environment
  • Benchmark performance of the algorithm in various scenarios

Department

Vanderlande Industries delivers material handling systems to control product flows in warehousing, sorting centers for package delivery, and baggage handling systems on airports.

Within Vanderlande, FLEET is responsible for developing solutions based on autonomous vehicle technology. These activities are carried out in close cooperation with other companies within Toyota Industries.

Your profile

  • Background and interest in algorithm design
  • Background and interest in performance optimization
  • Strong analytical skills

Vanderlande is the global market leader for value-added logistic process automation at airports, and in the parcel market. Vanderlande’s baggage handling systems move 4.2 billion pieces of luggage around the world per year. Its systems are active in 600 airports including 14 of the world’s top 20.

Transport & Logistiek
Veghel
6.000 medewerkers