Nokia APPRENTI Digital twinning road supervision system 

France 

544902952

The twin is a digital representation modelling the actual system . In our specific case, the twin must be a "system of systems" combining various specialized twins, including traffic flows (based on SUMO , CARLA and other proprietary options.

The twin can be used for various purposes:

• Offline optimization
  : twin is used after some form of calibration (based on measurement campaign and edition) to validate new functions. Work can be done offline, and the optimized functions are deployed in a second step.
• Online optimization
  : twin is put "in sync" with deployed real time probes, representing a form of real time calibration. One example is provided in [REF 1] for SUMO. Other internal options exist for CARLA. Main challenge is to ensure accuracy of digital "copy" of the scene, with deterministic/guaranteed delay and event fidelity. Then, real time processes can be performed on the digital copy, in the "real time/online optimization" space.

1. Coordination of several subsystems : an orchestration is required between the various simulators, probes, actuators… composing the digital twin instance. This orchestration should be able to adopt a declarative approach separating the concerns for defining scenarios spanning several simulators of same (e.g. several coordinated SUMO instances) or different (eg SUMO, CARLA, other proprietary simulators) kinds. One example of such approach is given in [REF 3]. A common API between simulators enabling atomic CRUD operations is required, possibly building on top of existing tools (eg SUMO TRACI). This work falls in the vast space of "co-simulation" and will benefit from preliminary internal assessments.
2. Interface with "scene" : a data platform facilitating interfaces with external scenes is under development. It enables to capture measurements from magnetic loops, ITS messages, video streams… the proposed orchestration needs to be able to ingest these inputs. For control space, only virtual controllers will be provided in the first case for security reasons. The quality/fidelity of the synchronization will happen at this stage and represents a major step, especially for use-cases relating to safety. Solutions will be needed to "harden" the fidelity (hot standby, stateless/microservice deployment on K8S clusters, system monitoring, bad functioning detection by cross checking monitored data based on ML…)
3. Use-case validation: validation will be performed on several spots in mixed reality (combining real and simulated measurements), selection will be done depending on each project progression. Some specific challenges use-cases will be selected from the proposed list and used as stress test. Successful demo will be shared with high profile customers from our ecosystem.

[REF 1] "Building a Motorway Digital Twin in SUMO: Real-Time Simulation of Continuous Data Stream from Traffic Counters" - Kresimir Kusic, René Schumann, Edouard Ivanjko

[REF 2] LLM-Assisted Light: Leveraging Large Language Model Capabilities for Human-Mimetic Traffic Signal Control in Complex Urban Environments - Maonan Wang, Aoyu Pang, Yuheng Kan, Man-On Pun, Chung Shue Chen and Bo Huang

[REF 3] Large Language Model Enhanced Multi-Agent Systems for 6G Communications- Feibo Jiang, Li Dong, Yubo Peng, Kezhi Wang, Kun Yang, Fellow, Cunhua Pan, Dusit Niyato, Octavia A. Dobre

• Carla and/or Sumo simulation environment: must have
• C/C++ and/or Python advanced: must have
• Stateless/full, distributed software architectures, asynchronous systems, "Cloud natives": must have
• object oriented/functional programming: preferable.

Nokia has received the following recognitions for its commitment to inclusion & equality:

• One of the World’s Most Ethical Companies by Ethisphere
• Gender-Equality Index by Bloomberg
• Workplace Pride Global Benchmark