Neuromorphic Robotic Lab
Neuromorphic Robotic Lab
Embedded Files
The Neuromorphic Robot Lab at Michigan Tech has state-of-the-art equipment and apparatus for studying neuromorphic robots in the fields of artificial intelligence, neuroscience, and cognition, focusing on aspects such as learning, memory, and decision-making. Our lab is fully equipped to conduct indoor experiments with robots, including obstacle avoidance, associative memory and learning, navigation, and more.
The neuormorphic robotic lab aims to utilize neuromorphic computing on mobile robots to realize a self-learning, real-time, and energy-efficient system. Specifically, the sensory signals of the robotics, such as Lidar, camera, etc., will be encoded into spiking signals for regular tasks, e.g., perception, learning, decision-making, and navigation.
Fear Conditioning Experimental Environment
Fear Conditioning Experimental Environment
The neuromorphic lab includes a novel fear conditioning experiment environment that uses vibration as an aversive stimulus and visual cues as neural stimuli. This setup is specifically designed to emulate and study the associative learning processes observed in rodents.
The fear conditioning experiment provides a controlled environment to investigate how neuromorphic robots can learn and adapt to adverse conditions through real-time interaction with their environment. This experimental setup holds substantial potential for studying fear responses and associative learning, offering insights that could inform both robotic control systems and our understanding of learning and memory processes in rodents.
T-maze Experimental Environment
T-maze Experimental Environment
The neuromorphic lab includes an innovative T-maze setup that utilizes neuromorphic robots as surrogates for rodents in decision-making experiments. The T-maze maintains its classic T-shape, providing a binary choice environment to assess learning, memory, and decision-making similar to traditional rodent experiments.
This setup supports various experimental protocols, including rewarded and spontaneous alternation tasks, and can be adapted for different stimuli types such as visual, auditory, and tactile cues.
Open-Field Experimental Environment
Open-Field Experimental Environment
The Neuromorphic Robot Lab at Michigan Tech features an advanced circular open field arena designed to replicate the experimental conditions used for studying memory, learning, and spatial behavior in rodents.
The arena's circular design, equipped with various markers and obstacles, allows robots to engage in associative learning, spatial memory, and decision-making experiments, closely emulating rodent behaviors in a similar circular arena. The flexibility of the arena setup supports a variety of experiments, including place preference tests, object recognition, and spatial navigation tasks, making it a versatile tool for cognitive and behavioral studies.
Traffic Simulation Environment
Traffic Simulation Environment
The Neuromorphic Robotics Lab is equipped with a traffic simulation table that enhances our research and educational endeavors in autonomous vehicles. This simulation table creates a realistic mini-city environment, complete with roads, traffic lights, stop signs, and various obstacles. It allows our neuromorphic robots to perform tasks such as precise self-localization, SLAM (Simultaneous Localization and Mapping), route planning, traffic light recognition, and autonomous obstacle avoidance.
Jackal is a small, fast, entry-level field robotics research platform. It has an onboard computer, GPS, and IMU fully integrated with ROS for out-of-the-box autonomous capability.
Our Jackal UGV is equipped with 3D Lidar, depth camera, IMU, NVIDIA Jetson Nano Developer Kit, etc.
The LIMO mobile robot is a multi-modal, ROS-powered robot development platform that integrates four motion modes: omnidirectional, tracked, Ackermann, and four-wheel differential. These modes are complemented by robust perception sensors and the power of the Nvidia Jetson Nano. This combination makes LIMO an optimal choice for advancing the development of diverse indoor and outdoor industrial applications while facilitating the exploration and implementation of ROS-based frameworks.
As a member of the Intel Neuromorphic Research Community (INRC), our group has full access to the computational resources, specifically to the most powerful neuromorphic server named Loihi-2, which deployed in fall 2021. Also, we have Kapoho Bay Loihi chips (left figure) with a USB interface that has 262 K neurons for our intelligent robot and the neuromorphic Deep Brain Stimulation (DBS) systems.
Our lab has multiple robot arms available for educational and research purposes. With different accessories, such as a suction cup, pneumatic gripper, and writing kits, the robotic arms can conduct multiple tasks, such as 3D printing, laser engraving, drawing, etc. The robotic arm supports secondary development in 13 interfaces and 20 programming languages/methods, such as Bluetooth, Wi-Fi, ROS, Python, PLC, Arduino, etc.
Our group also has GPU-based deep learning computational platforms for students as listed below:
2 of NVIDIA TITAN RTX 24GB GDDR6
5 of Intel Xeon Silver 24 Threads, 2.2 GHz 16.5 MB Cache, 85W, 4214 Processor, 12 cores
20 of NVIDIA Tesla V100 32GB CoWoS HBM2 PCIe 3.0
Parts of computational resources are supported by the Institution of Computing and Cybersystems (ICC) Group. The Institute of Computing and Cybersystems (ICC) is composed of six centers that promote research and learning experiences in the areas of cyber-physical systems, cybersecurity, data sciences, human-centered computing, and scalable architectures and systems for the benefit of Michigan Tech and society at large. The ICC has close to 60 faculty members and it provides research support through seed grants, travel awards, student funding, support personnel, lab infrastructure, etc. Laboratory infrastructure of note includes the Data Sciences Laboratory, the Immersive Visualization Studio, and Systems Control Lab. The ICC also provides a full-time research support staff for member faculty.
Prof. An is a faculty member of the Centers for Cyber-Physical Systems, Human-Centered Computing, and Scalable Architectures and Systems
More details can be found at https://www.mtu.edu/icc/centers/
Page updated
Report abuse