Repository der Technischen Hochschule Ingolstadt
Not a member yet
4170 research outputs found
Sort by
Fast Dynamic Difficulty Adjustment for Intelligent Tutoring Systems with Small Datasets
Full text linkThis paper studies the problem of automatically adjusting the difficulty level of educational exercises to facilitate learning. Previous work on this topic either relies on large datasets or requires multiple interactions before it adjusts properly. Although this is sufficient for large-scale online courses, there are also scenarios where students are expected to only work through a few trials. In these cases, the adjustment needs to respond to only a few data points. To accommodate this, we propose a novel difficulty adjustment method that requires less data and adapts faster. Our proposed method refits an existing item response theory model to work on smaller datasets by generalizing based on attributes of the exercises. To adapt faster, we additionally introduce a discount value that weakens the influence of past interactions. We evaluate our proposed method on simulations and a user study using an example graph theory lecture. Our results show that our approach indeed succeeds in adjusting to learners quickly
Document Knowledge Graph to Enhance Question Answering with Retrieval Augmented Generation
No full text
Infra-3DRC-FusionNet: Deep Fusion of Roadside Mounted RGB Mono Camera and Three-Dimensional Automotive Radar for Traffic User Detection
Full text linkMono RGB cameras and automotive radar sensors provide a complementary information set that makes them excellent candidates for sensor data fusion to obtain robust traffic user detection. This has been widely used in the vehicle domain and recently introduced in roadside-mounted smart infrastructure-based road user detection. However, the performance of the most commonly used late fusion methods often degrades when the camera fails to detect road users in adverse environmental conditions. The solution is to fuse the data using deep neural networks at the early stage of the fusion pipeline to use the complete data provided by both sensors. Research has been carried out in this area, but is limited to vehicle-based sensor setups. Hence, this work proposes a novel deep neural network to jointly fuse RGB mono-camera images and 3D automotive radar point cloud data to obtain enhanced traffic user detection for the roadside-mounted smart infrastructure setup. Projected radar points are first used to generate anchors in image regions with a high likelihood of road users, including areas not visible to the camera. These anchors guide the prediction of 2D bounding boxes, object categories, and confidence scores. Valid detections are then used to segment radar points by instance, and the results are post-processed to produce final road user detections in the ground plane. The trained model is evaluated for different light and weather conditions using ground truth data from a lidar sensor. It provides a precision of 92%, recall of 78%, and F1-score of 85%. The proposed deep fusion methodology has 33%, 6%, and 21% absolute improvement in precision, recall, and F1-score, respectively, compared to object-level spatial fusion output
Comprehensive Analysis of Float Current Behavior and Calendar Aging Mechanisms in Lithium-Ion Batteries
No full textAiming to quantify degradation currents from solid electrolyte interphase formation (ISEIgrowth) and gain of active lithium due to cathode lithiation (ICL), resulting from electrolyte decomposition, the float current behavior of lithium-ion batteries is investigated with different cathode materials. The float current, IFloat , represents the recharge current required to maintain the cell at a fixed potential during calendar aging. This current arises as lithium is irreversibly consumed at the anode or inserted into the cathode, shifting the electrode potentials. To account for the asymmetric response of the electrodes, a voltage-dependent scaling factor, SF, is introduced, derived from the slopes of the electrode-specific voltage curves. Using this factor in combination with measured float currents and capacity loss rates from check-up tests, ISEIgrowth and ICL is quantified at 30 °C across various float voltages. Although the SF and capacity data are limited to 30 °C, the model is extended to a range of 5–50 °C using only float current measurements. The results show that using capacity loss rates alone underestimate ISEIgrowth and that ICL, contributes significantly to the observed float current at elevated voltages, indicating that cathode lithiation plays an increasingly important role in high-voltage calendar aging
A Glimpse into the Future: An Inverse Soft Q-Learning's Soft Actor-Critic Approach for Pedestrian Path Prediction
No full text
