The general objective of the present project NANO-MIND is to develop an efficient technology for AI sensing,
processing and actuating systems based on the asynchronous real-time generation, communication, processing
and learning of events. The project will contribute by developing sensors providing an efficient event-based
representation of the signal, developing event-based processing and learning architectures and the corresponding
cmos-memristive low power computation-in-memory hardware implementations, and developing event-based
motor control strategies. The developed AI event-based technology will be integrated in several demonstrators
where the benefits of the technology can be benchmarked with solutions offered by other state-of-art event-
based as well as conventional technologies.
This general objective can be achieved by targeting the specific objectives, where the specific sub-project and
responsible researcher are indicated:
1. Development of a novel SCD binocular vision sensor and camera.
2. Development of an SCD hybrid vision/magnetic imaging sensor and probe.
3. Development of a binaural cochlea sensor based on SSP.
4. Generation of experimental datasets from event-driven sensors in different scenarios to be used as trainers for
the considered SNNs.
5. Development of learning algorithms for echolocation and voice commands classification.
6. Implementation of spike-based neural network architectures combining stereo-vision, binaural auditory and
magnetic imaging input signals.
7. Development of learning algorithms for the spike-based neural network architecture proposed in objective 4
for different classification tasks of static stimulus.
8. Extension of the learning and categorization principles of objectives 4 and 5 to dynamic stimulus, such us,
9. Extension of event-based motor-control algorithms for robotics.
10. Development of a hybrid CMOS-memristive low-power hardware accelerator module for the implementation
of the architectures and learning schemes proposed in objectives 5-7 .
11. Adaptation of robotic platforms to integrate the event-based hardware developed in NANO-MIND .
12. Develop smart sensing platforms able to combine stereo-vision, binaural-auditory as well as opto-magnetic
signals. In particular:
a) Smart sensing platform for high speed gesture recognition.
b) Robotic vehicle to target a sound source and avoiding high-speed obstacles.
c) ED stimulated pocket drone.
d) High-speed NDT smart platform.
12+1. Demonstrate the performance of the developed smart sensing and robotic platforms in practical
applications comparing them with the state-of-the art solutions.