Situational awareness in intelligence is not only science, but also art (that is, individualized representation). The binocular parallax makes the stereo vision, and the SA difference between the human and the machine forms the holographic SA. State is calculation, potential is calculation, sense is mapping, and knowledge is connection. One of the difficulties in the research of situation awareness lies in: the occlusion relationship between situation and situation, the occlusion relationship between sense and knowledge, the occlusion relationship between saliency and value, and the occlusion relationship between fact and value! Yin-yang is not a contradiction often said in the West. Eastern thought has fewer contradictions and more Yin-yang. The difference between the East and the West lies in the understanding of being and should. West: being leads to should, and east: should leads to being. The game is like a student's exam. On the surface, it is a calculation of the question by a person, but in essence it is a calculation problem between the student and the person who made the question. Perhaps there is no such thing as a yuan in everything in the world. Even if there is a yuan, it is likely to be a variable, not a fixed yuan. Yuans also have a life cycle, constantly being generated and disappearing in interaction. This is the representation, the judgment, the cognition, and the intelligence. Everything is changing in motion. If you must find a coordinate origin as the “yuan” to deal with the current problem, then the “change” is the yuan? ! But for this, the “machine” that adapts to changes should not be a machine, but a “timing”. At present, only people can grasp the timing of complexity—those who are truly valuable!
The reason why the dark green system fails to be expected is not in the calculation of machines, but in human calculations. Calculations are often based on facts, and calculations often rely on value. Just like the relationship between labor prices and value in economics, in a game What opposes the interaction between the two sides is also objective factual data (may be called formalization here) and subjective value experience (may be called intentionality here), one concrete and the other abstract. In human-machine fusion intelligence, the speed and accuracy of machine calculations within a certain range can use deterministic formal facts to help people analyze, judge, and make decisions (just like prices can reflect value within a certain range, as in daily life). Vegetables, radishes, green onions), but some machine calculations that go beyond the uncertainty of the facts are powerless (just like the price cannot reflect the value in an abnormal range, such as the firewood, rice, oil and salt tickets in the epidemic), and people, especially those with relevant experience The value of calculations and stubbornness is revealed. Machine learning can extract some relevant habitual preference feature values based on objective factual data, but these are conditional and limited values, involving designers, users, and other temporal and spatial contexts, and the boundaries of imagination logic. The real value emerges more. Situational awareness with height, depth and even temperature is the product of mutual interaction between intention and form, sensibility and reason, subjective and objective. Therefore, the failure of Deep Green is due to the deviation of fact and value, the separation of subjective and objective, and the error of calculation and calculation.
Intelligence also partially obeys the following three axioms (unfortunately, these assumptions are not implemented in many places in the dark green system):
1. Axioms of similarities and differences: There are two sets of transformations between any objective things. One kind of transformation makes two things broadly equivalent, and the other kind of transformation makes two kinds of things unequal.
2. Connection axiom 1 (things and connections) For any objective thing ai, there are other things aj, ak that interact with ai. They form a certain connection R (aj, ak......).
3. Connection axiom 2 (things and connections) For the connection Ra between arbitrarily selected things, there is always another connection Rb, so that Ra belongs to Rb or Rb belongs to Ra.
Appendix: "dark green" (DeepGreen, DG) is the US Defense Advanced Research Projects Agency (DARPA) research development of the next generation of operational command and decision support system to " observe - judge - Decision - Action " ( the OODA ) loop The " observation - judgment " link of the company demonstrated the possible effects of using different combat plans through multiple computer simulations, predicting the enemy’s actions, allowing the commander to make correct decisions, and shortening the time required to formulate and analyze combat plans. Time, take the initiative to deal with the enemy instead of passively coping with the attack, so that the US military commander can be one step ahead of potential opponents in thought and action.
Figure 1 Schematic diagram of the principle of the dark green system
The " Dark Green " system is mainly composed of a human-computer interaction module named " Commander Assistant " , a simulation module named " Blitzkrieg " , and a decision generation module named " Crystal Ball " . Its architecture is shown in Figure 2 .
Figure 2 Dark green system architecture diagram
(1) "Commander Assistant" module.
The "Commander Assistant" module mainly completes the man-machine dialogue function, which can automatically convert the commander's hand-drawn sketches and the corresponding language expressing command intentions into brigade-level action plans (COA), helping to quickly generate combat plans and make quick decisions. This module includes the following 3 sub-modules: "planning sketch", "decision sketching" and "automatic plan generation":
● "Planning Sketch" sub-module.
"Planning Sketch" has the following functions: accepting user's hand-drawn sketches and voice input, and transforming them into standard military symbols, such as the US military combat symbol guidelines MILSTD2525B. Commanders can think and draw in their own way, instead of sticking to a completely formal military standard; add details to the combat plan; have enough knowledge in various fields, when encountering a few unclear questions, they can ask the user to understand Real intent and initialize the combat model.
The output of the "plan sketch" will be an action plan described in military scenario markup language. "Planning sketch" includes sketch recognizer, plan inducer, plan automatic generator, detail adding planner and dialogue generator. The sketch recognizer converts a series of self-drawn marks and voices into a series of standard military symbols; the plan inducer uses a large number of symbols to help the commander make plans and intentions; the detail addition planner will add details to the plan generated by the commander, so The "blitzkrieg" module can simulate the program; the dialogue generator can interact with the commander to clarify ambiguities and help understand the commander's determination and intentions.
● "Decision Drafting" sub-module.
The “decision sketch” is critical to achieving the “dark green” goal. Its purpose is to enable the commander to “see the future”. It has the following functions: receiving input from the “crystal ball” decision points and decisions from the commander; displaying different decisions The multi-dimensional information such as the possibility, risk, value, effect and other factors generated by the plan helps the commander better understand the situation that may be formed in the future; and communicates the decision to the subordinate.
The "decision sketch" includes a probe module, a presentation module, a dialog generator, and a command generator. The exploration module allows the commander to explore possible future combat images, so as to grasp the follow-up effect of the decision; the presentation module converts the information from the future combat image into an intuitive representation; the dialog generator presents the required decision to the commander and communicates with the commander Communicate until the commander’s combat intention is truly understood; the command generator expresses the commander’s decision-making specifications as instructions to his subordinates, and provides this information to the "crystal ball" module for maintaining and updating future combat images.
● "Automatic scheme generation" sub-module
In the early stage of the "Dark Green" program, the "automatic plan generation" sub-module simply transformed the commander's intention into a combat plan. With the advancement of the "Dark Green" plan, the goal of this module is to creatively and automatically generate a combat plan that meets the commander's intentions.
(2) "Blitzkrieg" module
The "blitzkrieg" module is the simulation part of the "dark green" plan. By using qualitative and quantitative analysis tools, various decision-making plans proposed by the commander can be quickly simulated to generate a series of possible future results. This module has a self-learning function, and the ability to predict future results can be continuously improved.
The "blitzkrieg" module can identify each decision branch point, thereby predict the range and possibility of possible results, and then follow each decision path to simulate. The "blitzkrieg" module mainly includes three parts: a multi-decision simulator, a model and behavior library, and a geospatial database. It has the following functions: inputting the plans of all parties in the war; determining the decision branch point or possible future situation; reasoning and evaluating each decision branch Possibility; continuous simulation of all decisions, traversing all possible decision choices.
(3) "Crystal Ball" Module
The "Crystal Ball" module will be able to make more accurate predictions of the future combat process in a timely manner based on the information in the combat process. Its main functions include: in the process of generating possible future results, it receives the decision-making plan from the "planning sketch", and then sends it to the "blitzkrieg" module for simulation, and then receives feedback from the "blitzkrieg" module, and converts it in a quantitative form. Perform a comprehensive analysis of all possible future results; obtain updated information from ongoing combat operations, and at the same time update the possibility parameters of various possible future results; use these updated possibility parameters to analyze and compare possible future results to The commander provides the most likely future results; uses the analysis results to determine the upcoming decision point, reminds the commander to make further decisions, and calls the "decision sketch."