Papers
15
Total Citations
229
H-Index
10
About
Alec Koppel is a researcher whose work spans reinforcement learning, distributed robotics, machine learning optimization, and probabilistic inference — areas where theoretical rigor meets real-world autonomous systems. His most impactful contribution, the ASAPP algorithm for asynchronous distributed pose graph optimization (2020, 45 citations), broke new ground in multi-robot SLAM by eliminating the synchronization bottlenecks that previously hampered collaborative mapping. Equally influential is his theoretical analysis of policy gradient methods (2019, 44 citations), which helped establish formal global convergence guarantees for a class of reinforcement learning algorithms widely used in robotics and gaming. Koppel has also made significant advances in scalable machine learning, pioneering doubly stochastic successive convex approximation methods for large-scale dictionary learning and nonconvex optimization. His work on scalable Gaussian process regression addresses a fundamental computational bottleneck in Bayesian nonparametric inference, with applications in robotics and chemical engineering. A recurring theme across his portfolio is making sophisticated probabilistic and optimization methods viable for decentralized, resource-constrained multi-agent systems — a contribution increasingly relevant as autonomous robot teams become central to real-world deployment scenarios.
Research Focus
Key Achievements
Top Papers
- 1Asynchronous and Parallel Distributed Pose Graph Optimization45 citations · 2020
- 2
- 3D4L: Decentralized Dynamic Discriminative Dictionary Learning22 citations · 2017
- 4
- 5
- 6
- 7
- 8D4L: Decentralized dynamic discriminative dictionary learning13 citations · 2015
- 9
- 10