Andreas Seefried

Dr. Andreas Seefried

orcid

Research Focus & Metrics

Academic Career Length

11Years (2016-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
18 March 2024 in IEEE Transactions on Vehicular Technology
Verified
Journal Article
Closed access
HSHendrik Scheidel
Hendrik Scheidel
Houshyar AsadiHoushyar Asadi
Houshyar Asadi
TBTobias Bellmann
Tobias Bellmann
ASAndreas Seefried
Andreas Seefried
SMShady Mohamed
Shady Mohamed
SNSaeid Nahavandi
Saeid Nahavandi
In the field of motion simulation, the level of immersion strongly depends on the motion cueing algorithm (MCA), as it transfers the reference motion of the simulated vehicle to a motion of the motion simulation platform (MSP). The challenge for the MCA is to reproduce the motion perception of a real vehicle driver as accurately as possible without exceeding the limits of the workspace of the MSP in order to provide a realistic virtual driving experience. In case of a large discrepancy between the perceived motion signals and the optical cues, motion sickness may occur with the typical symptoms of nausea, dizziness, headache and fatigue. Existing approaches either produce non-optimal results, e.g. due to filtering, linearization, or simplifications, or the required computational time exceeds the real-time requirements of a closed-loop application. This work presents a new solution to the motion cueing problem, where instead of a human designer specifying the principles of the MCA, an artificial intelligence (AI) learns the optimal motion by trial and error in interaction with the MSP. To achieve this, a well-established deep reinforcement learning (RL) algorithm is applied, where an agent interacts with an environment formulated as a Markov decision process (MDP). This allows the agent to directly control a simulated MSP to obtain feedback on its performance in terms of platform workspace usage and the motion acting on the simulator user. The RL algorithm used is proximal policy optimization (PPO), where the value function and the policy corresponding to the control strategy are learned and both are mapped in artificial neural networks (ANN). This approach is implemented in Python and the functionality is demonstrated by the practical example of pre-recorded lateral maneuvers. The subsequent validation on a standardized double lane change shows that the RL algorithm is able to learn the control strategy and improve the quality of the immersion compared to an established method while enhancing the realistic driving motion sensation. Thereby, both the perceived translational accelerations and rotational angular velocities determined under consideration of the vestibular system are more accurately reproduced, and the resources of the MSP are used more economically.

1 January 2018 in Engineering Applications of Artificial Intelligence
Verified
Journal Article
Closed access
TLThomas Lombaerts
Thomas Lombaerts
GLGertjan Looye
Gertjan Looye
ASAndreas Seefried
Andreas Seefried
MNMiguel Neves
Miguel Neves
TBTobias Bellmann
Tobias Bellmann

1 April 2019
Verified
Conference Paper
Closed access
ASAndreas Seefried
Andreas Seefried
APAlexander Pollok
Alexander Pollok
RKRichard Kuchar
Richard Kuchar
MHMatthias Hellerer
Matthias Hellerer
MLMartin Leitner
Martin Leitner
TBTobias Bellmann
Tobias Bellmann


21 December 2019 in at - Automatisierungstechnik
Verified
Journal Article
Closed access
PDF file
ASAndreas Seefried
Andreas Seefried
APAndreas Pfeiffer
Andreas Pfeiffer
ZusammenfassungDer Beitrag stellt einen Algorithmus zur Generierung von mehrfach stetig differenzierbaren Trajektorien auf Basis von B-Splines vor, deren Funktionswerte und Ableitungen innerhalb von konstanten Grenzen liegen. Die Trajektorien können für Systeme genutzt werden, bei denen Grenzen in den Ableitungen des Systemeingangs vorliegen. Dies trifft beispielsweise auf Modelle zu, bei denen ein Antrieb mit Begrenzungen in der Position, Geschwindigkeit und Beschleunigung als Systemeingang genutzt wird.Zur Generierung werden Parameter, Startwerte der Trajektorie sowie die konstanten Grenzen benötigt. Eine quadratische Optimierung liefert die Stützpunkte der B-Spline-Kurve, die dadurch definiert ist. Die implizit begrenzten B-Splines können ideal in Optimalsteuerungsproblemen und modellprädiktiven Regelungen als Steuertrajektorie eingesetzt werden. Durch die implizite Berücksichtigung von Nebenbedingungen können ebensolche aus dem eigentlichen Optimierungsproblem entfallen. Daraus resultiert eine Verringerung der Komplexität des Optimalsteuerungsproblems bzw. der modellprädiktiven Regelung was wiederum zu einer schnelleren Berechnungszeit führt.Die implizit begrenzten B-Splines werden im Beitrag vorgestellt und exemplarisch in einer modellprädiktiven Regelung mit einem robotischen System angewandt. Sie führen zu einer deutlich schnelleren Berechnungszeit im Vergleich zur Nutzung von B-Splines mit expliziten Nebenbedingungen im Optimierungsproblem.

18 April 2016
Verified
Conference Paper
Open access
ASAndreas Seefried
Andreas Seefried
APAndreas Pfeiffer
Andreas Pfeiffer
In this work-in-progress paper, a currently ongoing development of a generic tool for nonlinear model predictive control is presented. By using an extended interface of FMI 2.0, it is possible to simulate a model that acts as prediction model while the actual system is simulated simultaneously. A trajectory optimization that uses the prediction model provides optimized input control values for the actual system at every sample time. The current work is based on the Optimization library for Dymola and an extended version of FMI 2.0 Co-Simulation. The structure of this approach is explained in detail as well as possible settings and limitations. An example shows the practicability and an outlook for further development is given.