Truncated Approximate Dynamic Programming with Task-Dependent Terminal Value
Authors: Amir-massoud Farahmand, Daniel Nikovski, Yuji Igarashi, Hiroki Konaka
AAAI 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We also empirically validate this approach in a real-world application of designing an energy management system for Hybrid Electric Vehicles with promising results. 4 Experiments We apply the proposed method to design an energy management system for a hybrid vehicle (Sciarretta and Guzzella 2007). |
| Researcher Affiliation | Industry | Amir-massoud Farahmand and Daniel N. Nikovski Mitsubishi Electric Research Laboratories Cambridge, MA, USA Yuji Igarashi and Hiroki Konaka Mitsubishi Electric Corporation Hyogo 661-8661, Japan |
| Pseudocode | Yes | Algorithm 1 Truncated ADP Solver with Task-Dependent Terminal Value Algorithm 2 TRUNCATEDADP(Rθ, ˆPθ, t, H, Vθ) |
| Open Source Code | No | The paper does not provide any statement or link regarding the public availability of its source code. |
| Open Datasets | No | The paper states: 'We generate several random run-curves θi, find the optimal value function V θi by dynamic programming, and select several random points in time and state from each random run-curve. These points define the training data.' However, it does not provide concrete access information (link, DOI, repository, or formal citation) for this dataset. |
| Dataset Splits | No | The paper mentions 'number of training data points' and 'generating a new random run-curve' for evaluation, but does not provide specific details on training, validation, or test dataset splits (e.g., percentages, sample counts, or predefined citations). |
| Hardware Specification | No | The paper mentions 'onboard computers' and 'a powerful computer of the car manufacturer' in a general context but does not provide specific details about the hardware (e.g., GPU/CPU models, memory) used for running the experiments. |
| Software Dependencies | No | The paper mentions using a 'reproducing kernel Hilbert space (RKHS)-based regularized least-squares regression estimator' and 'DP (with 2000 discretization of the state variable x)', but does not list specific software dependencies with version numbers (e.g., Python, PyTorch, TensorFlow, or specific solver versions). |
| Experiment Setup | Yes | We set T to 115 with each time step corresponding to one minute of physical time, so we plan for an almost two hour trip. The parameter θ is a vector of dimension 230 describing the altitude and the speed profile of the run-curve. Instead of directly working with a 230 + 2-dimensional vector (the vector of θ as well as x and t), we extract a 7-dimensional features that approximately summarize the route and the current state of the vehicle. We then use these features as an input to a reproducing kernel Hilbert space (RKHS)-based regularized least-squares regression estimator to find the estimate of V . We use DP (with 2000 discretization of the state variable x) and perform H iterations of the Bellman backup. |