Table 1.
Learning rules based on STDP/Hebbian learning.
| Learning rule | Robot | Sensor | Methodology | Reference |
|---|---|---|---|---|
| Unsupervised | Two-wheel vehicle | 5 Proximity Sensors | Implementing an SNN on a resistive memory device and apply it to navigation tasks | Sarim et al., 2016a,b |
| Mobile vehicle Casis-I | 16 Ultrasonic Sensors | A behavior-based target-approaching navigation controller composed of three sub-controllers: the obstacle-avoidance, wall-following, and goal-approaching SNN controllers. | Wang et al., 2014, 2008 | |
| TriBot Robot | Distance Sensor, Contact Sensor, | Using SNN to make robot navigate in an unknown environment and avoid obstacles | Arena et al., 2010 | |
| Supervised | Two-wheel insect | 4 Proximity Sensors | Implementing an indirect training SNN in digital CMOS to navigate with obstacles | Hu et al., 2014; Mazumder et al., 2016 |
| Two-wheel insect | 2 Terrain, 2 Target | Indirectly train an SNN by RBFs to determine precise spike timings and minimize a desired objective function | Zhang et al., 2013 | |
| Aircraft | IMU | Indirectly training an SNN to approximate an optimal flight controller | Foderaro et al., 2010 | |
| 4-DoF robotic arm | 4 Joint Encoder, 3 Spatial direction of end-effector | Using supervised learning to train a single-layer network to control a robotic arm with 4 degrees of freedom in 3D space | Bouganis and Shanahan, 2010 | |
| 2-DoF robotic arm | Sensorimotor | Using supervised learning to train a spiking model of the cerebellum to control a robotic arm | Carrillo et al., 2008 | |
| STDP / Hebbian Learning | ||||
| Conditioning | Simulated fly | Olfactory Receptor | Implementing an SNN inspired by Drosophila olfactory system to simulate flight | Faghihi et al., 2017 |
| Lego EV3 robotic platform | Camera, Infrared sensor Colour/light sensor | Learning and unlearning autonomously locomotion based on visual-input with reinforced/aversive reflex-response | Jimenez-Romero, 2017 | |
| Two-wheel robot | 3 Proximity Sensors, 1 RGB Sensor | Using Reward-dependent STDP learning rule to allow OC and CC learning | Dumesnil et al., 2016a,b | |
| Foraging Ants | Olfactory Sensors, Nociceptor | Learns to associate olfactory sensor input with different behaviors through a single-layer SNN | Jimenez-Romero et al., 2015, 2016 | |
| Lego NXT 2.0 | Color sensor, Touch Sensor | Using SNN to sustain OC in multiple learning scenarios | Cyr et al., 2014; Cyr and Thériault, 2015 | |
| Two-wheel Vehicle | Light Sensors | Using light sensors in a target-reaching task to punish wrongful behavior | Iwadate et al., 2014 | |
| Two-wheel Vehicle | 5 Proximity Sensors, 9 IR Sensors, Vibration Sensor | Using infrared, ultrasound and visual neurons as CS and vibration neurons as US | Cyr and Boukadoum, 2012 | |
| Mobile vehicle Casis-I | 16 Ultrasonic Sensors | A learning algorithm combining operant conditioning and a shunting neural dynamics model is applied to the path planning | Wang et al., 2012 | |
| TriBot Robot | Distance Sensor, Camera, Contact Sensor, | Using target distance as CS, while contact sensors work as US causing an unconditioned response | Arena et al., 2009a,b | |
| R-STDP | Flapping Insect | GPS and IMU | Indirectly training an SNN-based controller for adaptive flight control | Clawson et al., 2016 |
| 1-DoF robotics arm | 5 Proximity Sensors | Using an SNN trained by a global reward and punishment signal to reach arbitrary targets | Spüler et al., 2015 | |
| Musculoskeletal arm, WAM robot | Encoders | Using a cortical spiking model composed of several hundred spiking model-neurons to control a two-joint arm | Dura-Bernal et al., 2015 | |
| CARL-SJR | Tactile Sensors | Using SNN to provide feedback to users by displaying bright colors on its surface. | Chou et al., 2015 | |
| Two-wheel vehicle | 2 Proximity Sensors | Implement a version of DA-modulated STDP on a food foraging task | Evans, 2015 | |
| Foraging Simulator | Visual Sensors | Using reward-STDP based SNN to solve a grid-based foraging task | Skorheim et al., 2014 | |
| DfRobotShop Rover | Camera, Light Sensor | Using an SNN and external flash to reinforce the goal-directed and adaptive behaviors | Helgadottir et al., 2013 | |
| 2-DoF robotics arm | Sensorimotor | Using an SNN based on R-STDP to control a two-joint virtual arm to reach to a fixed target | Neymotin et al., 2013 | |
| 1-DoF robotics arm | Encoder | Using an SNN to control a single-joint arm for target reaching | Chadderdon et al., 2012 | |