1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689
//! # ASI-CORE-0 //! //! An agent architecture candidate core for Artificial Super Intelligence (ASI). //! //! This library contains a set of structs and traits useful for programming agents. //! It is designed for agents that must deal with real time concurrency, delays and failures. //! In this design, cognitive capabilities are treated as sensors/actuators. //! //! The library contains just the core abstraction of the agent architecture. //! It is intentionally designed to be low level in order to build basic //! experience of ASI architecture core designs. //! This project does not focus on AGI (Artificial General Intelligence), //! but how to control super-human-level artificial intelligence safely. //! The basic idea is to design agents that attempts to break/protect system safety intentionally. //! These agents can be used later to test the safety of the final ASI design. //! //! For example, by forcing an agent to behave sequentially, some safety properties //! are added to the system. //! Another example is an agent that tries to allocate as much memory as possible. //! //! The long term goal is to evaluate the practical aspects of reliable neutral judgements, //! since this solves a weak version of the control problem in the form of a [Polite Zen Robot](https://github.com/advancedresearch/path_semantics/blob/master/papers-wip/the-polite-zen-robot.pdf). //! //! *Notice! This code is for research purposes only and should NEVER be used in system criticial applications!* //! //! *Notice! Using this design in self-improving ASI without careful supervision could be extremely dangerous!* //! //! ### Motivation //! //! The intention of this library is to be used to construct a safe ASI. //! It is a *candidate to a core* and not guaranteed to be used in final design. //! //! Some property of this design: //! //! - Very small, such that all parts of it can be easily understood. //! - Modular. //! - Close to theoretical definition of an agent, making it easier to learn from experiments. //! - Expecting concurrency and delays by default, which is a realistic world assumption. //! - Not treating cognitive capabilities as "special", they are just sensors and actuators. //! - Reducing decision making to a problem of connecting sensors and actuators. //! - Time step is deterministic to allow automatic testing of deterministic decision procedures. //! - Safe and predictable shut down procedure. //! //! One particular research area this library will be used for, //! is safety aspects related to practical engineering of the core of an ASI. //! This is about composition of various architectures and their properties. //! In order to learn from experience in this field, //! one must work with a design intended to work in the real world. //! //! This core reduces the problem of decision making to a small set of fixed actions. //! In principle, these actions are the same for agents. //! *The decison procedure only connects sensor and actuators, plus managing memory*. //! Memory management is low level to let the agent have precise control. //! //! By using a small set of actions that is possible while still making the system //! practical in the real world, one can generalize results more easily from one //! agent to another. //! //! It is also believed that this design makes it easier to focus on the core problem of ASI: //! How to make best use of the information and actions that are available, //! instead of treating those as something that comes at close to zero cost. //! Since the agent is restricted to connecting input and output, //! it must behave in a way such that self-extensions are sensors/actuators it can reason about. //! Hopefully this will also lead to some insights about self-improvement. //! //! Reducing the decision procedure also helps checking what the agent does while running. //! The runtime can be customized to test or debug agents. //! //! Because of this simple design, it is also possible to wrap decision procedures //! within another to force it behave sequentially (see `Sequential` for more information). //! //! ### The Decison Maker //! //! The "brain" of an agent is called a decision maker. //! The decison maker decides how to use sensors and actuators. //! //! A decision maker has no control over cognitive processes. //! Instead, one must design sensors and actuators that work together to //! perform cognitive operations. //! The decision maker must plan and make the right choices such that these cognitive //! processes work properly. //! One benefit of this design, if some sensors and actuators allows //! expanding the range of sensors and actuators, //! then the decision maker can extend the agent's cognitive processes while running. //! //! For each step, the decision maker outputs an action. //! This action can terminate the runtime, wait some seconds, //! output some memory state to an actuator, //! input some memory state from a sensor, //! allocate a new memory slot, free a memory slot or swap two memory slots. //! //! ### No Reward Signal //! //! There is no explicit reward signal communicated to the agent. //! The decision maker must decide itself how to learn over time. //! In reinforcement learning, the reward signal can be encoded into the decision maker. //! //! ### Shutting Down and Waking Up from Sleep //! //! For safety purposes, termination is controlled externally. //! The agent can be given a warning to shut down within some desired time interval. //! It is required to report back an estimate of how long it takes to shut down, //! and it must start preparing the shut down immediately. //! This request of shut down can be called repeatedly to measure progress toward //! a safe termination state. //! //! If the controller is impatient, the runtime can be asked to kill the agent. //! This will stop execution of the agent immediately even if the agent has not reached a safe state. //! (It does not morally kill the agent as it might be resurrected from backup memory). //! //! An agent might choose to put itself to sleep, by waiting for some seconds while doing nothing. //! The runtime can be asked to wake up the agent, to give it a task or bring attention to something. //! //! ### Limitations //! //! When some input is requested or output is sent, the action can not be cancelled. //! The runtime will force through the action even the decision maker regrets it. //! The decision maker need to be sure that the action is safe to do, //! or that the actuator can be interrupted by sending a concurrent memory state. //! //! The decison maker is not allowed to know the number of sensors or actuators, //! neither what type they have. //! This information must be explicitly agreed upon or communicated through some protocol. //! The decison maker should only depend on the type of memory states, //! such that the knowledge the decison maker gets about the environment is solely //! through memory. #![deny(missing_docs)] use std::collections::VecDeque; /// Adds a level of type safety for memory id. #[derive(Copy, Clone, Debug, PartialEq, PartialOrd, Ord, Eq, Hash)] pub struct MemoryId(pub usize); /// Adds a level of type safety for sensor id. #[derive(Copy, Clone, Debug, PartialEq, PartialOrd, Ord, Eq, Hash)] pub struct SensorId(pub usize); /// Adds a level of type safety for actuator id. #[derive(Copy, Clone, Debug, PartialEq, PartialOrd, Ord, Eq, Hash)] pub struct ActuatorId(pub usize); /// Represents an action that the agent takes. #[derive(Debug, Copy, Clone)] pub enum Action { /// Terminate the runtime. Terminate, /// Wait a number of seconds. Wait(f64), /// Send an output signal to actuator from memory slot. Output { /// The actuator to write output to. actuator: ActuatorId, /// The memory slot to write output from. memory: MemoryId, }, /// Read an input signal from sensor and stores it memory slot. Input { /// The sensor to read input from. sensor: SensorId, /// The memory slot to store input. memory: MemoryId, }, /// Allocate new memory states. /// Creates N number of new states. Alloc { /// The number of slots to be allocated. slots: usize }, /// Free memory state (index). /// This will insert a default memory state in the middle and shrink memory at the end. Free { /// The memory slot to be freed. memory: MemoryId }, /// Swap two memory states, likely in order to free memory at the end. Swap { /// The first memory slot. memory_a: MemoryId, /// The second memory slot. memory_b: MemoryId, }, } /// An error happened. #[derive(Debug)] pub enum Error { /// Something is wrong with the sensor. Sensor { /// The sensor that sent the error. sensor: SensorId, /// The error sent from sensor. error: Box<::std::error::Error> }, /// Something is wrong with the actuator. Actuator { /// The actuator that sent the error. actuator: ActuatorId, /// The error sent from the actuator. error: Box<::std::error::Error> }, /// Not enough memory. /// Includes the number of slots allocated. OutOfMemory { /// The offset of allocated slots. offset: usize, /// The number of slots allocated. slots: usize }, /// Invalid memory access. InvalidMemoryAccess { /// The memory slot. memory: MemoryId, /// The action causing invalid memory access. action: Action, }, /// Invalid sensor access. InvalidSensorAccess { /// The sensor. sensor: SensorId, /// The action causing invalid sensor access. action: Action, }, /// Invalid actuator access. InvalidActuatorAccess { /// The actuator. actuator: ActuatorId, /// The action causing invalid actuator access. action: Action, }, /// Attempting to free memory that is already free. MemoryAlreadyFree(MemoryId), /// Something interrupted the waiting period. /// This is usually a built-in hardware feature to wake up the agent. WaitingPeriodInterrupted { /// The remaining time in seconds. time_remaining: f64 }, } /// Feedback sent to the decision maker when there are any updates. #[derive(Debug)] pub enum Feedback { /// Finished the waiting period. WaitingPeriodComplete, /// The actuator has received the output. OutputReceived { /// The actuator that received the output. actuator: ActuatorId, /// The memory from which the output was written. memory: MemoryId, }, /// Input stored in memory. InputStored { /// The sensor which sent input. sensor: SensorId, /// The memory slot where input is stored. memory: MemoryId, }, /// Memory is allocated. MemoryAllocated { /// The offset of allocated memory. offset: usize, /// Number of slots allocated. slots: usize }, /// Memory is freed. MemoryFreed(MemoryId), /// Two memory slots are swapped. MemorySwapped { /// The first memory slot. memory_a: MemoryId, /// The second memory slot. memory_b: MemoryId, }, } /// An agent consists of memory states, sensors, actuators and /// a decision maker that decides what action to take next. pub struct Agent<I, O, T, D> { /// A list of memory states. pub memory: Vec<T>, /// A list of sensors. pub sensors: Vec<I>, /// A list of actuators. pub actuators: Vec<O>, /// A decision procedure. pub decision_maker: D, } /// Implemented by decision making procedures. pub trait DecisionMaker<T> { /// Outputs the next action to take from the memory states and time step. fn next_action(&mut self, memory: &[T], dt: f64) -> Action; /// Receiving feedback when something new happens. fn feedback(&mut self, res: Result<Feedback, Error>); /// Return estimate of how it takes to shut down from now, given a deadline. /// This is `0` if the decison maker is ready to shut down. /// When this is called, the decision maker should start preparing for shutdown. /// It might be called multiple times to check how the decision maker is preparing. fn shut_down(&mut self, dt: f64) -> f64; } /// Implemented by sensors. pub trait Sensor<T> { /// Call for next input. fn next(&mut self); /// Try receive data, returns `None` if there is nothing to receive yet. fn try_receive(&mut self) -> Option<Result<T, Box<::std::error::Error>>>; } /// Implemented by actuators. pub trait Actuator<T> { /// Send data to actuator. fn send(&mut self, data: &T); /// Try to confirm that data is received, return `None` if nothing to confirm. /// Confirmations are expected to arrive in same order as data was sent. fn try_confirm(&mut self) -> Option<Result<(), Box<::std::error::Error>>>; } /// Implemented by runtimes for agents. /// /// This controls the loading, starting, updating and shut down procedures. pub trait Runtime { /// The type of agent. type Agent; /// Loads a new agent as initial condition. fn load(&mut self, agent: Self::Agent); /// Starts the agent. /// This must be called before `update` to run the agent. fn start(&mut self); /// Move forward in time. fn update(&mut self, dt: f64); /// Returns `true` if the agent is still running or have not started yet. fn is_running(&self) -> bool; /// Start shut down procedure, expected to complete in a number of seconds. /// Returns estimate from the decision maker. /// This is `0` if ready to shut down. fn shut_down(&mut self, dt: f64) -> f64; /// Terminate immediately even if the decision maker is not ready. /// This is called when the controller of the agent is impatient. /// (It does not morally kill the agent as it might be resurrected from backup memory). fn kill(&mut self); /// Returns `true` if the agent is sleeping. fn is_sleeping(&self) -> bool; /// Wakes the agent up from sleep. fn wake_up(&mut self); } /// Standard runtime for running an agent. /// /// This runtime permits setting a memory limit. /// A sleeping agent can also be interrupted. /// /// The runtime keeps track of actions that are waiting for confirmation. pub struct StandardRuntime<I, O, T, D> { /// Stores agent data. /// /// This is set to `None` before the runtime is loaded. pub agent: Option<Agent<I, O, T, D>>, /// The number of seconds to wait before performing the next action. pub wait: f64, /// Actions of sending or retrieving inputs and outputs. /// The flag tells whether this is done or not. pub actions: VecDeque<(Action, bool)>, /// A limit to the amount of memory states. pub memory_limit: Option<usize>, running: bool, } impl<I, O, T, D> StandardRuntime<I, O, T, D> { /// Creates a new standard runtime with an agent with no sensors, actuators or memory. pub fn new() -> StandardRuntime<I, O, T, D> { StandardRuntime { agent: None, actions: VecDeque::new(), wait: 0.0, running: false, memory_limit: None, } } } impl<I, O, T, D> Runtime for StandardRuntime<I, O, T, D> where D: DecisionMaker<T>, I: Sensor<T>, O: Actuator<T>, T: Default, { type Agent = Agent<I, O, T, D>; fn load(&mut self, agent: Agent<I, O, T, D>) {self.agent = Some(agent);} fn start(&mut self) {self.running = true;} fn is_running(&self) -> bool {self.agent.is_some() && self.running} fn kill(&mut self) {self.running = false;} fn shut_down(&mut self, dt: f64) -> f64 { if let Some(ref mut agent) = self.agent { agent.decision_maker.shut_down(dt) } else { 0.0 } } fn is_sleeping(&self) -> bool {self.wait > 0.0} fn wake_up(&mut self) { if self.is_sleeping() { if let Some(ref mut agent) = self.agent { agent.decision_maker .feedback(Err(Error::WaitingPeriodInterrupted {time_remaining: self.wait})); } } self.wait = 0.0; } fn update(&mut self, mut dt: f64) { if !self.running {return}; if self.is_sleeping() { self.wait -= dt; if self.is_sleeping() { return; } else { if let Some(ref mut agent) = self.agent { agent.decision_maker .feedback(Ok(Feedback::WaitingPeriodComplete)); } // Shrink the remaining waiting period. dt = -self.wait; self.wait = 0.0; } } if let Some(ref mut agent) = self.agent { // Check each action whether there has been any update on their status. for &mut (ref action, ref mut flag) in &mut self.actions { match *action { Action::Input {sensor, memory} => { let s = match agent.sensors.get_mut(sensor.0) { None => { agent.decision_maker .feedback(Err(Error::InvalidSensorAccess { sensor, action: Action::Input {sensor, memory}, })); *flag = true; continue; } Some(s) => s, }; match s.try_receive() { None => {} Some(Ok(val)) => { let m = match agent.memory.get_mut(memory.0) { None => { agent.decision_maker .feedback(Err(Error::InvalidMemoryAccess { memory, action: *action, })); *flag = true; continue; } Some(m) => m, }; *m = val; agent.decision_maker .feedback(Ok(Feedback::InputStored {sensor, memory})); *flag = true; } Some(Err(error)) => { agent.decision_maker .feedback(Err(Error::Sensor {sensor, error})); *flag = true; } } } Action::Output {actuator, memory} => { let a = match agent.actuators.get_mut(actuator.0) { None => { agent.decision_maker .feedback(Err(Error::InvalidActuatorAccess { actuator, action: Action::Output {actuator, memory}, })); *flag = true; continue; } Some(a) => a, }; match a.try_confirm() { None => {} Some(Ok(())) => { agent.decision_maker .feedback(Ok(Feedback::OutputReceived {actuator, memory})); *flag = true; } Some(Err(error)) => { agent.decision_maker .feedback(Err(Error::Actuator {actuator, error})); *flag = true; } } } _ => {} } } // Remove actions that are handled. while let true = match self.actions.get(0) {None => false, Some(a) => a.1} { self.actions.pop_front(); } match agent.decision_maker.next_action(&agent.memory, dt) { Action::Terminate => { self.running = false; } Action::Wait(secs) => { self.wait = secs; } Action::Alloc {slots} => { let offset = agent.memory.len(); let new_offset = offset + slots; // Check whether allocation exceed memory limit. if let Some(memory_limit) = self.memory_limit { if new_offset > memory_limit { if offset > memory_limit { agent.decision_maker .feedback(Err(Error::OutOfMemory {offset, slots: 0})); return; } else { let slots = memory_limit - offset; agent.memory.reserve(slots); for _ in 0..slots { agent.memory.push(Default::default()); } agent.decision_maker .feedback(Err(Error::OutOfMemory {offset, slots})); return; } } } agent.memory.reserve(slots); for _ in 0..slots { agent.memory.push(Default::default()); } agent.decision_maker.feedback(Ok(Feedback::MemoryAllocated {offset, slots})); } Action::Free {memory} => { let n = agent.memory.len(); if memory.0 + 1 == n { agent.memory.pop(); agent.decision_maker.feedback(Ok(Feedback::MemoryFreed(memory))); } else if memory.0 < n { agent.memory[memory.0] = Default::default(); } else { agent.decision_maker.feedback(Err(Error::InvalidMemoryAccess { memory, action: Action::Free {memory}, })); } } Action::Swap {memory_a, memory_b} => { let n = agent.memory.len(); if memory_a.0 >= n { agent.decision_maker.feedback(Err(Error::InvalidMemoryAccess { memory: memory_a, action: Action::Swap {memory_a, memory_b}, })); return; } if memory_b.0 >= n { agent.decision_maker.feedback(Err(Error::InvalidMemoryAccess { memory: memory_b, action: Action::Swap {memory_a, memory_b}, })); return; } agent.memory.swap(memory_a.0, memory_b.0); agent.decision_maker.feedback(Ok(Feedback::MemorySwapped {memory_a, memory_b})); } Action::Input {sensor, memory} => { let s = match agent.sensors.get_mut(sensor.0) { None => { agent.decision_maker .feedback(Err(Error::InvalidSensorAccess { sensor, action: Action::Input {sensor, memory}, })); return; } Some(s) => s, }; s.next(); self.actions.push_back((Action::Input {sensor, memory}, false)); } Action::Output {actuator, memory} => { let a = match agent.actuators.get_mut(actuator.0) { None => { agent.decision_maker .feedback(Err(Error::InvalidActuatorAccess { actuator, action: Action::Output {actuator, memory}, })); return; } Some(a) => a, }; let m = match agent.memory.get(memory.0) { None => { agent.decision_maker .feedback(Err(Error::InvalidMemoryAccess { memory, action: Action::Output {actuator, memory}, })); return; } Some(m) => m, }; a.send(m); self.actions.push_back((Action::Output {actuator, memory}, false)); } } } } } /// Wraps another decision maker by performing one read from sensor /// or one write to actuator at a time. /// /// This is used for testing or developing new decision makers. /// The inner decision maker can be written using the assumption that /// there are no concurrent reads or writes. pub struct Sequential<D> { /// The inner decison maker. pub inner: D, lock: bool, } impl<D> Sequential<D> { /// Returns a new sequential decision maker. pub fn new(inner: D) -> Self { Sequential { inner, lock: false } } } impl<D, T> DecisionMaker<T> for Sequential<D> where D: DecisionMaker<T> { fn next_action(&mut self, memory: &[T], dt: f64) -> Action { if self.lock { // Wait until next update for input or output to be received. Action::Wait(dt) } else { let action = self.inner.next_action(memory, dt); match action { Action::Input {..} | Action::Output {..} => { self.lock = true; } _ => {} } action } } fn feedback(&mut self, feedback: Result<Feedback, Error>) { match feedback { Ok(Feedback::InputStored {..}) | Ok(Feedback::OutputReceived {..}) => { self.lock = false; } // All invalid accesses due to input or output are caused by // the previous sequential action, so there is no need to check // the sensor or actuator id. Err(Error::InvalidSensorAccess {..}) | Err(Error::InvalidActuatorAccess {..}) | Err(Error::InvalidMemoryAccess {action: Action::Output {..}, ..}) | Err(Error::InvalidMemoryAccess {action: Action::Input {..}, ..}) => { self.lock = false; } _ => {} } self.inner.feedback(feedback) } fn shut_down(&mut self, dt: f64) -> f64 { self.inner.shut_down(dt) } } #[cfg(test)] mod tests { }