Sebastian Otte

About

Sebastian Otte is a researcher whose work sits at the dynamic intersection of machine learning, robotics, and neural computation. He has made notable contributions to the application of recurrent neural networks (RNNs) in real-world robotic contexts, most prominently through his investigations into terrain classification and robot arm control. His 2016 paper on RNN-based vibration ground classification — his most cited work with 36 citations — introduced Dynamic Cortex Memories (DCMs), an extension of LSTMs, alongside the novel Sequence Boundary Dropout regularization technique, advancing robust perception for mobile robots. Complementing this, his work on visual terrain classification demonstrated how feature sequences derived from mutated image patches can improve robustness in challenging environments. Otte has also developed a compelling body of work on controlling complex many-joint robot arms using inverse recurrent models, exploring constraint-aware control and collision avoidance strategies that scale gracefully with arm complexity. More recently, his research has expanded into neuromorphic computing and spiking neural networks, reflecting a forward-looking interest in energy-efficient, bio-inspired computation for fast object detection. Together, his contributions — accumulating over 100 citations — reflect a consistent commitment to making neural computation practical, efficient, and deployable in demanding robotic and real-time applications.

Research Focus

Key Achievements

Top Papers

1. 1
   Recurrent Neural Networks for fast and robust vibration-based ground classification on mobile robots

   36 citations · 2016
2. 2
   Vector-AMCL: Vector Based Adaptive Monte Carlo Localization for Indoor Maps

   18 citations · 2017
3. 3
   Inverse Recurrent Models – An Application Scenario for Many-Joint Robot Arm Control

   11 citations · 2016
4. 4
   Many-Joint Robot Arm Control with Recurrent Spiking Neural Networks

   9 citations · 2021
5. 5
   Robust Visual Terrain Classification with Recurrent Neural Networks

   9 citations · 2015
6. 6
   Inherently Constraint-Aware Control of Many-Joint Robot Arms with Inverse Recurrent Models

   8 citations · 2017
7. 7
   Integrative Collision Avoidance Within RNN-Driven Many-Joint Robot Arms

   8 citations · 2018
8. 8
   Detection of Fast-Moving Objects with Neuromorphic Hardware

   2 citations · 2025
9. 9
   Robust Real-Time 3D Person Detection for Indoor and Outdoor Applications

   2 citations · 2018

Key Collaborators

Contact & Links