Information Technology Reference
In-Depth Information
all afferent inputs, and by means of a three-step process (i.e. data-dimensionality
reduction, features extraction and classification) permits an holistic impression of the
external environment. This would be the starting point for intelligent devices to
perceive the external world instead of simply sensing it.
Another important aspect that emerges from the current analysis regards so-called
embodiment. This is the idea that biological systems contain something that is called
structural intelligence or morphological computation. In other words, the morphology
of biological systems is naturally oriented towards performing specific tasks. In this
way, biological systems can reduce the effort needed to control their motion. Among
others, a critical example is the muscle skeletal apparatus of humans. In fact, the
antagonistic muscles powering a joint, because of their intrinsic non-linear elastic
properties, make the articulation intrinsically stable and easily controlled.
The combination of sensory fusion and a morphology with computational skill would
increase the intelligence level of the device. The idea of defining the roadmap priorities
in the short, middle and long term is to gradually decrease the computational effort
dedicated to motion control (i.e. the system will have a morphological computation
capability) while increasing the efficiency and the effectiveness of the sensory fusion
techniques (i.e. the system will be able to perceive the external world holistically):
investigation and development of reliable and efficient sensory fusion strategies;
strengthening of morphological computation design paradigm (2015); development of
platforms integrating both advanced sensory fusion strategies and having
morphological computation skills (2020); investigation and development of complex
high-level strategies using the sensory fusion strategy to promote adaptive perception-
action mechanisms (2025).
Mechatronic robotic devices (robotic hand)
An interesting field where the concept of sensor and actuator integration has been
exploited is that of robot hands for prosthetics. In fact, in such a field the need for low
power, low weight while still retaining “dexterous and sensorized” fingers is of
paramount importance. Several examples of intrinsic prosthetic hands in research may
be listed: the Southampton-REMEDI hand, the RTR II hand, the MANUS hand and the
Karlsruhe hand. The SmartHand prosthesis (Controzzi, 2008), developed by SSSA
within the PRIN2006 program is an innovative transradial hand because of its tight
design that includes actuators, a control system and an extensive sensory system with
40 sensors. Many other examples may be listed: the TBM hand (Dechev, 2001), the
RTR II hand (Massa, 2002), the Soft hand (Carrozza, 2005), the KNU hand (Chu,
2008). Other significant research related to extrinsic hands used as bionic prostheses
platforms include the Cyberhand (Carrozza, 2006, Cipriani, 2008b), the Yokoi hand
(Ishikawa, 2000), and the Vanderbilt University prototypes (Fite, 2008).
Exploiting RFID technology, sensors will include transceiver units for wireless
communication with the host controller. The reduction of wired buses will drastically
increase the robustness of devices. Innovative low-power consumption buses for such
communication systems will then be available. The development of smaller actuators
with high efficiency will allow engineers to fit high numbers of sensors and actuators
inside intelligent devices.
Sensor technology will be strongly dependent on silicon development. In 2020
sensors will include programmable digital-signal processing algorithms and filters
for the autonomous extraction of its features. Signal processing together with
wireless transmission will reduce the burden of the host controller dealing with such
