Accelerating Robot Dynamics Gradients on a CPU, GPU, and FPGA
Computing the gradient of rigid body dynamics is a central operation in many state-of-the-art planning and control algorithms in robotics. Parallel computing platforms such as GPUs and FPGAs can offer performance gains for algorithms with hardware-compatible computational structures. In this letter,...
| Main Authors: | , , , , , |
|---|---|
| Other Authors: | |
| Format: | Article |
| Language: | English |
| Published: |
Institute of Electrical and Electronics Engineers (IEEE)
2022
|
| Online Access: | https://hdl.handle.net/1721.1/143469 |
| _version_ | 1826198325059125248 |
|---|---|
| author | Plancher, Brian Neuman, Sabrina M Bourgeat, Thomas Kuindersma, Scott Devadas, Srinivas Reddi, Vijay Janapa |
| author2 | Massachusetts Institute of Technology. Computer Science and Artificial Intelligence Laboratory |
| author_facet | Massachusetts Institute of Technology. Computer Science and Artificial Intelligence Laboratory Plancher, Brian Neuman, Sabrina M Bourgeat, Thomas Kuindersma, Scott Devadas, Srinivas Reddi, Vijay Janapa |
| author_sort | Plancher, Brian |
| collection | MIT |
| description | Computing the gradient of rigid body dynamics is a central operation in many state-of-the-art planning and control algorithms in robotics. Parallel computing platforms such as GPUs and FPGAs can offer performance gains for algorithms with hardware-compatible computational structures. In this letter, we detail the designs of three faster than state-of-the-art implementations of the gradient of rigid body dynamics on a CPU, GPU, and FPGA. Our optimized FPGA and GPU implementations provide as much as a 3.0x end-to-end speedup over our optimized CPU implementation by refactoring the algorithm to exploit its computational features, e.g., parallelism at different granularities. We also find that the relative performance across hardware platforms depends on the number of parallel gradient evaluations required. |
| first_indexed | 2024-09-23T11:03:05Z |
| format | Article |
| id | mit-1721.1/143469 |
| institution | Massachusetts Institute of Technology |
| language | English |
| last_indexed | 2024-09-23T11:03:05Z |
| publishDate | 2022 |
| publisher | Institute of Electrical and Electronics Engineers (IEEE) |
| record_format | dspace |
| spelling | mit-1721.1/1434692023-02-06T15:41:50Z Accelerating Robot Dynamics Gradients on a CPU, GPU, and FPGA Plancher, Brian Neuman, Sabrina M Bourgeat, Thomas Kuindersma, Scott Devadas, Srinivas Reddi, Vijay Janapa Massachusetts Institute of Technology. Computer Science and Artificial Intelligence Laboratory Computing the gradient of rigid body dynamics is a central operation in many state-of-the-art planning and control algorithms in robotics. Parallel computing platforms such as GPUs and FPGAs can offer performance gains for algorithms with hardware-compatible computational structures. In this letter, we detail the designs of three faster than state-of-the-art implementations of the gradient of rigid body dynamics on a CPU, GPU, and FPGA. Our optimized FPGA and GPU implementations provide as much as a 3.0x end-to-end speedup over our optimized CPU implementation by refactoring the algorithm to exploit its computational features, e.g., parallelism at different granularities. We also find that the relative performance across hardware platforms depends on the number of parallel gradient evaluations required. 2022-06-17T16:43:26Z 2022-06-17T16:43:26Z 2021 2022-06-17T16:39:13Z Article http://purl.org/eprint/type/JournalArticle https://hdl.handle.net/1721.1/143469 Plancher, Brian, Neuman, Sabrina M, Bourgeat, Thomas, Kuindersma, Scott, Devadas, Srinivas et al. 2021. "Accelerating Robot Dynamics Gradients on a CPU, GPU, and FPGA." IEEE Robotics and Automation Letters, 6 (2). en 10.1109/LRA.2021.3057845 IEEE Robotics and Automation Letters Creative Commons Attribution-Noncommercial-Share Alike http://creativecommons.org/licenses/by-nc-sa/4.0/ application/pdf Institute of Electrical and Electronics Engineers (IEEE) MIT web domain |
| spellingShingle | Plancher, Brian Neuman, Sabrina M Bourgeat, Thomas Kuindersma, Scott Devadas, Srinivas Reddi, Vijay Janapa Accelerating Robot Dynamics Gradients on a CPU, GPU, and FPGA |
| title | Accelerating Robot Dynamics Gradients on a CPU, GPU, and FPGA |
| title_full | Accelerating Robot Dynamics Gradients on a CPU, GPU, and FPGA |
| title_fullStr | Accelerating Robot Dynamics Gradients on a CPU, GPU, and FPGA |
| title_full_unstemmed | Accelerating Robot Dynamics Gradients on a CPU, GPU, and FPGA |
| title_short | Accelerating Robot Dynamics Gradients on a CPU, GPU, and FPGA |
| title_sort | accelerating robot dynamics gradients on a cpu gpu and fpga |
| url | https://hdl.handle.net/1721.1/143469 |
| work_keys_str_mv | AT plancherbrian acceleratingrobotdynamicsgradientsonacpugpuandfpga AT neumansabrinam acceleratingrobotdynamicsgradientsonacpugpuandfpga AT bourgeatthomas acceleratingrobotdynamicsgradientsonacpugpuandfpga AT kuindersmascott acceleratingrobotdynamicsgradientsonacpugpuandfpga AT devadassrinivas acceleratingrobotdynamicsgradientsonacpugpuandfpga AT reddivijayjanapa acceleratingrobotdynamicsgradientsonacpugpuandfpga |