Periodic Reporting for period 1 - NI (Natural Intelligence for Robotic Monitoring of Habitats)
Periodo di rendicontazione: 2021-01-01 al 2022-06-30
The European Union answer to Global Warming is the set of deeply transformative policies contained in the European Green Deal. Therein, a prominent role is given to monitoring, restoring, and preserving the ecosystems of the Natura 2000 Network. Undoubtedly, today this is one of the top priorities for humanity.
Today, monitoring of such a large area can be performed only by human operators. The reason is not only their expertise and knowledge in classifying plants and habitats, but also because their physical intelligence allows them to move for hours in wild unstructured environments, such dunes, forests, and mountains. However, humans not always guarantee repeatable, consistent, and affordable monitoring in order to assess the conservation status of ecosystems in order to timely act towards Nature preservation.
Instead, robotics can theoretically allow such monitoring, but, despite tremendous advancements in recent years, robots hardly leave laboratories and factories since they are not robust and efficient to survive in the real world. This main issue is due to the lack of autonomy of flying robots and also since ground robots are not: i) intelligent to autonomously percept, interpret, and interact with highly uneven, slipping and irregular grounds, ii) physically robust to manage unexpected contacts and impacts.
The Natural Intelligence (NI) project aims to serve the European Green Deal via monitoring the natural habitats of N2000N with robots able to effectively move in dunes, grasslands, forests, and alpine terrains. Natural Intelligence will enable artificial Cognition for a) autonomous classification of plants species and natural habitats, b) autonomous navigation in natural environments, and c) effective physical environment-robot interaction through environment-aware impedance planning and bio-inspired anticipatory control. Within NI, articulated soft-robotics powered mechatronics will a) enable enhanced locomotion and terrain perception using adaptive bio-inspired feet, b) improve robustness through novel robust-by-design articulated soft robot structures, and c) allow long-lasting operation capabilities through efficient exploitation of robot dynamics.
NI project will provide as main outputs:
• the first examples of robotic workforce for monitoring natural habitats (to TRL 4 and above) with system prototype demonstrations in real environments: forests, grasslands, dunes, and alpine scenarios;
• pre-standards and guidelines making the results of NI available at a decision-making level that will pave the way for the adoption of a robotic workforce for environmental monitoring putting Europe in a worldwide leading position;
To this end, the project envisions ambitious research and innovation activities focused on objectives related to: science and technology, usability, and standardization.
Today, monitoring of such a large area can be performed only by human operators. The reason is not only their expertise and knowledge in classifying plants and habitats, but also because their physical intelligence allows them to move for hours in wild unstructured environments, such dunes, forests, and mountains. However, humans not always guarantee repeatable, consistent, and affordable monitoring in order to assess the conservation status of ecosystems in order to timely act towards Nature preservation.
Instead, robotics can theoretically allow such monitoring, but, despite tremendous advancements in recent years, robots hardly leave laboratories and factories since they are not robust and efficient to survive in the real world. This main issue is due to the lack of autonomy of flying robots and also since ground robots are not: i) intelligent to autonomously percept, interpret, and interact with highly uneven, slipping and irregular grounds, ii) physically robust to manage unexpected contacts and impacts.
The Natural Intelligence (NI) project aims to serve the European Green Deal via monitoring the natural habitats of N2000N with robots able to effectively move in dunes, grasslands, forests, and alpine terrains. Natural Intelligence will enable artificial Cognition for a) autonomous classification of plants species and natural habitats, b) autonomous navigation in natural environments, and c) effective physical environment-robot interaction through environment-aware impedance planning and bio-inspired anticipatory control. Within NI, articulated soft-robotics powered mechatronics will a) enable enhanced locomotion and terrain perception using adaptive bio-inspired feet, b) improve robustness through novel robust-by-design articulated soft robot structures, and c) allow long-lasting operation capabilities through efficient exploitation of robot dynamics.
NI project will provide as main outputs:
• the first examples of robotic workforce for monitoring natural habitats (to TRL 4 and above) with system prototype demonstrations in real environments: forests, grasslands, dunes, and alpine scenarios;
• pre-standards and guidelines making the results of NI available at a decision-making level that will pave the way for the adoption of a robotic workforce for environmental monitoring putting Europe in a worldwide leading position;
To this end, the project envisions ambitious research and innovation activities focused on objectives related to: science and technology, usability, and standardization.
In the first period of the project - from month 1 to month 18 – the carried out technical activities show that the project is on track, and no major deviations need to be reported.
Remarkably, the main point of the project – that is, to prove that robotic technology is ready to be used in real environmental monitoring use-case to acquire data usable to determine the habitat health status - has been accomplished in (more than) five field missions in the four habitat of the project in spots belonging to the Natura 2000 Network.
A labeled database is under development - accounting for about 1000 annotated items - thanks to the collaboration between roboticists, data scientists, and botanists.
Benchmarking activities, conducted on the basis of the field observations, have been able to predict the robot behavior in several use cases and now provide tests and facilities able to validate system performances.
Robotics research activities resulted in the following:
- novel hardware to build highly dynamic, adaptive, and perceptive robot bodies;
- novel algorithms to let robot minds understand the traversability of an (unknown) environment, plan footholds taking into account feet functionality, plan recovery strategies from falls, and plan and control interactive motion able to exploit robot compliance.
As explained in the description reported in the following of this document the first two milestones of the project have been successfully reached: MS1 - Requirements - M6 - Reports with inputs for the design of technological components, use-case scenarios and existing standards. MS2 - Preliminary Integrator I2 - M6 - preliminary prototype of integrator I2 and preliminary prototypes of each component of integrator I3.
Major issues during the project have been related to the covid-19 pandemic scenario (lockdowns, lack of materials, and possible delays), however, the mitigation measures allowed the consortium to carry on the activities without major delays.
Remarkably, the main point of the project – that is, to prove that robotic technology is ready to be used in real environmental monitoring use-case to acquire data usable to determine the habitat health status - has been accomplished in (more than) five field missions in the four habitat of the project in spots belonging to the Natura 2000 Network.
A labeled database is under development - accounting for about 1000 annotated items - thanks to the collaboration between roboticists, data scientists, and botanists.
Benchmarking activities, conducted on the basis of the field observations, have been able to predict the robot behavior in several use cases and now provide tests and facilities able to validate system performances.
Robotics research activities resulted in the following:
- novel hardware to build highly dynamic, adaptive, and perceptive robot bodies;
- novel algorithms to let robot minds understand the traversability of an (unknown) environment, plan footholds taking into account feet functionality, plan recovery strategies from falls, and plan and control interactive motion able to exploit robot compliance.
As explained in the description reported in the following of this document the first two milestones of the project have been successfully reached: MS1 - Requirements - M6 - Reports with inputs for the design of technological components, use-case scenarios and existing standards. MS2 - Preliminary Integrator I2 - M6 - preliminary prototype of integrator I2 and preliminary prototypes of each component of integrator I3.
Major issues during the project have been related to the covid-19 pandemic scenario (lockdowns, lack of materials, and possible delays), however, the mitigation measures allowed the consortium to carry on the activities without major delays.
Natural Intelligence emerges through the interaction of three elements: environment, body, and mind. Hence, NI will develop through the following:
1) Natural-environment tailored methods leveraging on-board sensing and machine learning tools to autonomously classify plants species and habitats, achieved in WP5;
2) Action and motion planning and control algorithms for environment-aware, efficient, robust, and autonomous behavior in natural habitats, goals of WP3 and WP4;
3) Robot bodies based on soft-robotics technologies inspired by vertebrate animals, to enable highly dynamic and efficient motions, such as jumping and hopping; hard physical environment-robot interactions; and fall resiliency (WP2 and WP6);
4) Adaptive foot designs to enable locomotion and improve perception of state-of-the-art and novel quadruped robots in different, extreme environments such as dunes, grasslands, forests, and alpine scenarios (achieved in WP2).
NI will also
1) Challenge and demonstrate the contribution of NI to boost the robotic habitat monitoring application in realistic use cases. We will explicitly steer the project developments through a careful evaluation of the technologies in realistic demonstrations by considering real world use-cases. NI has strong use-case driven nature and four end-users - UNISI, UNIMI, UNISS, and UNIPG – will provide four different use-cases for field validation: dunes, grasslands, forests, and alpine environments (WP7);
2) Devise benchmarking approaches, methods, protocols, and facilities to assess the robot capability to move in natural habitats (WP1).
Several will be the public and private sectors that will benefit from the NI research achievements, relevant for environmental and socio-economic reasons:
- Development and validation of a dependable system for routine inspection of natural habitat, will substantially help European and National public bodies to match the ambitious target to monitor the 30% of European land by 2030.
- The robotization of monitoring tasks will improve autonomy, robustness, and efficiency of robots that can be directly exploited in the whole inspection and maintenance, agri-food, and search and rescue application areas.
- Strengthening Europe’s position in the global marketplace through support of SMEs and commercialisation of developed technology into new markets.
1) Natural-environment tailored methods leveraging on-board sensing and machine learning tools to autonomously classify plants species and habitats, achieved in WP5;
2) Action and motion planning and control algorithms for environment-aware, efficient, robust, and autonomous behavior in natural habitats, goals of WP3 and WP4;
3) Robot bodies based on soft-robotics technologies inspired by vertebrate animals, to enable highly dynamic and efficient motions, such as jumping and hopping; hard physical environment-robot interactions; and fall resiliency (WP2 and WP6);
4) Adaptive foot designs to enable locomotion and improve perception of state-of-the-art and novel quadruped robots in different, extreme environments such as dunes, grasslands, forests, and alpine scenarios (achieved in WP2).
NI will also
1) Challenge and demonstrate the contribution of NI to boost the robotic habitat monitoring application in realistic use cases. We will explicitly steer the project developments through a careful evaluation of the technologies in realistic demonstrations by considering real world use-cases. NI has strong use-case driven nature and four end-users - UNISI, UNIMI, UNISS, and UNIPG – will provide four different use-cases for field validation: dunes, grasslands, forests, and alpine environments (WP7);
2) Devise benchmarking approaches, methods, protocols, and facilities to assess the robot capability to move in natural habitats (WP1).
Several will be the public and private sectors that will benefit from the NI research achievements, relevant for environmental and socio-economic reasons:
- Development and validation of a dependable system for routine inspection of natural habitat, will substantially help European and National public bodies to match the ambitious target to monitor the 30% of European land by 2030.
- The robotization of monitoring tasks will improve autonomy, robustness, and efficiency of robots that can be directly exploited in the whole inspection and maintenance, agri-food, and search and rescue application areas.
- Strengthening Europe’s position in the global marketplace through support of SMEs and commercialisation of developed technology into new markets.