About

Gerardo G. Acosta is a leading researcher at the intersection of autonomous robotics, adaptive control, and underwater systems, whose work has profoundly shaped how intelligent machines navigate and interact with complex environments. His most celebrated contributions lie in the application of reinforcement learning to robot control — most notably his development of adaptive deep reinforcement learning frameworks for MIMO PID control of mobile robots (147 citations) and his incremental Q-learning strategy for adaptive PID control (99 citations), which together established practical, data-driven pathways to autonomous robot tuning. His Double Q-PID algorithm (78 citations) further refined these methods, demonstrating consistent innovation across a decade of research. Beyond terrestrial robotics, Acosta has made significant strides in autonomous underwater vehicle (AUV) navigation, including end-to-end deep reinforcement learning for AUV position tracking and novel sonar-based perception systems — particularly his CA-CFAR approach to sidescan sonar object detection (87 citations) — addressing real-world offshore monitoring demands. His earlier work on evolutionary neurocontrollers and artificial potential fields rounds out a remarkably broad portfolio. More recently, his ROS-based digital twin framework signals a forward-looking pivot toward smart manufacturing, cementing his relevance across multiple frontiers of autonomous systems research.

Research Focus

Key Achievements

13
H-Index
23
Papers
734
Total Citations
32
Avg Citations/Paper
🏆 Most Cited Paper
An adaptive deep reinforcement learning approach for MIMO PID control of mobile robots
147 citations · 2020
📈 Most Prolific Year: 2019 (4 Papers)
🤝 Key Collaborators: 28
🏛 Institutions: Consejo Nacional de Investigaciones Científicas y Técnicas, Universidad Nacional del Centro de la Provincia de Buenos Aires, Centro Científico Tecnológico - San Juan

Top Papers

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10

Key Collaborators

Contact & Links

Available for collaboration
Content generated · 14 days ago