CN103702413B - Real-time following and positioning system of indoor camera based on Zigbee network - Google Patents

Abstract

A real-time following and positioning system of an indoor camera based on a Zigbee network comprises a positioning RFID (Radio Frequency Identification) tag for marking a target, an RFID reader-writer module connected with the positioning RFID tag, a Zigbee network environment for wireless positioning, and a camera holder for controlling camera following, wherein the RFID reader-writer module comprises an antenna, an RFID chip, a microcontroller and a Zigbee signal transmit-receive module; the Zigbee network environment comprises the Zigbee network and a positioning mater controller; the Zigbee network comprises a Zigbee wireless network consisting of a Zigbee network coordinator and Zigbee network nodes; the positioning mater controller comprises a communication interface module for receiving a Zigbee network environment signal, an embedded information processing platform and a camera holder executing mechanism control module; and the embedded information processing platform integrates an echo state network positioning algorithm based on an RSSI (Received Signal Strength Indicator) model. The system can solve the problems of lack of a special positioning system, a large positioning error, difficult real-time positioning and the like of the existing indoor camera.

Description

Alignment system is followed in real time based on the live pick up head of zigbee network

Technical field

The invention belongs to wireless sensor network positioning technical field, specifically a kind of live pick up based on zigbee network Head follows alignment system in real time.

Background technology

Live pick up head positioning has stronger application background, the especially real-time following feature of photographic head.At present, in room Interior photographic head follows positioning field in real time does not also have the feasible system of complete set and scheme, is more to rely on human assistance Realize.Current also someone's reference gps (global position system) localization method, the method is passed through embedded Gps module positions.But the location equipment being embedded into gps module is often existed and is easily lost due to indoor signal, or mobile supply Electricity limited and cannot normal work problem so that gps implement tracing and positioning lost efficacy.Meanwhile, live pick up head adopts gps module When, also have such problems as that application cost is higher, integration complexity is high, a small range limited precision and real-time be not high.

Content of the invention

The technical problem to be solved be exactly provide a kind of live pick up head based on zigbee network in real time with With alignment system, can solve the problem that existing live pick up head lacks proprietary alignment system, position error is big, real-time positioning is difficult etc. Problem.

For solving above-mentioned technical problem, a kind of the following in real time based on the live pick up head of zigbee network of present invention offer Alignment system, including the positioning rfid label for indicating target, the rfid reader/writer module being attached thereto, is used for carrying out wirelessly The zigbee network environment of positioning includes sky with the photographic head head controlling photographic head to follow movement, described rfid reader/writer module Line, rfid chip, microcontroller and zigbee signal transceiver module, described zigbee network environment includes zigbee network and determines Position main controller, the zigbee that described zigbee network includes being made up of Zigbee network coordinator and zigbee network node is wireless Network, described positioning main controller include for receive zigbee network ambient signal communication interface modules, at embedded information Platform and photographic head head actuating mechanism controls module；

The described embedded information processing platform is integrated with the echo state network location algorithm based on rssi model, according to fixed Wireless signal strength when position rfid label moves under zigbee network environment, merges, and it is fixed to be reduced by network training mode Position mean square error, thus be accurately positioned rfid label；Meanwhile, photographic head head actuating mechanism controls module is believed according to location data Breath, sends instruction dynamic adjustment head steering wheel；；

Described rssi model is:

$\mathrm{pl}\left(d\right)=\mathrm{pl}\left(d_0\right)-10\mathrm{nlg}\left(\frac{d}{d_0}\right)+\mathrm{\ϵ}---\left(1\right)$

In formula (1), d is the distance to beaconing nodes for the calculative unknown node, and n is path loss index, depends on Zigbee network environment, the signal intensity that pl (d) receives for unknown node, pl (d₀) it is reference distance d₀The signal intensity at place, d₀For the distance of reference mode to beaconing nodes, ε is position error；

Described echo state network location algorithm comprises the following steps:

1) sample phase: a number of Zigbee network coordinator or zigbee network are set in positioning region indoors Node as reference mode, after the completion of Zigbee network coordinator and zigbee network node networking, believe by the position of these nodes Breath all it is known that the mutual decay intensity rssi value of measurement, builds w_out, described w_outFor reserve pool dr(dynamic Reservior) it is the connection weight matrix inputting and exporting for output；This data is sent out by Zigbee network coordinator via uart Deliver to register control；Positioning rfid label is moved into the original state for network defined in zigbee network, that is, network is first Beginning state is 0, i.e. x (0)=0, and (u (n), n=1,2 ... are p) through input connection weight matrix w for training positioning sample_inIt is added into storage In standby pond dr, it is sequentially completed system mode and error outputCalculating and collection；Calculate output connection weight matrix w_out, Zigbee network coordinator starts to collect internal state variable from a certain moment m, and with vectorial (x₁(i),x₂(i)…,x_n(i)) (i=m, m+1 ... are p) that (p-m+1, n), corresponding sample location data ε (n) is also collected row composition matrix β simultaneously, and structure Become column vector τ (p-m+1,1)；

2) the weight computing stage: collect alignment system state matrix and positioning rfid label movement according in sample phase Sample data, calculates output connection weight matrix w_out；Due to state variable x (n) and outputBetween be linear relationship, and need Target to be realized is using echo state network reality outputApproach desired output ε (n), that is,

$\mathrm{\ϵ}\left(n\right)\≈\widehat{\mathrm{\ϵ}}\left(n\right)={\mathrm{\σ}}_{i=1}^t{w}_i^{\mathrm{out}}{x}_i\left(n\right)---\left(2\right)$

Namely wish to calculate weights(For matrix w_outElement), meet network positions mean square deviation minimum it may be assumed that

$\min \frac{1}{p-m+1}\underset{n=m}{\overset{p}{\mathrm{\σ}}}{(\mathrm{\ϵ}\left(n\right)-{\mathrm{\σ}}_{i=1}^l{w}_i^{\mathrm{out}}{x}_i\left(n\right))}^2---\left(3\right)$

Can be in the hope of the inverse matrix of matrix β according to formula (3), i.e. w_out=β^-1τ, so far, esns network training completes, training What good network can be directly used for positioning rfid label follows positioning in real time.

In technique scheme, the core wireless module of described Zigbee network coordinator and zigbee network node is equal Using cc2530 chip.

In technique scheme, described photographic head head actuating mechanism controls module, controlled by adjusting pwm mode Head turns to.

In technique scheme, described zigbee signal transceiver module and zigbee network environment adopt 2.4ghz radio frequency Signal.

In technique scheme, described positioning main controller is using double arm kernel processes units, described communication interface modules Using asynchronous receiving-transmitting transmitter.

Compared with prior art, the beneficial effects of the present invention is: 1) cost performance of the present invention is high, zigbee chip with respect to Bluetooth, wifi and gps chip are all more cheap；2) node based on zigbee chip for the network environment that the present invention builds, all There are route and two kinds of functions of main frame, using low complex degree and MANET mode, the topological structure of energy quick response existing network Change, and the new network topology structure of reconstruct of precise and high efficiency；3) energy-conserving and environment-protective of the present invention, have the characteristics that low in energy consumption, two sections 5 Under number battery electric power thus supplied, as long as the sustainable operation for months of zigbee chip；4) present invention is highly reliable, and network environment is stable, Safe；5) present invention, in the environment based on zigbee network, using RF identification (rfid) technology, and adopts based on rssi The echo state network location algorithm of model, overcomes traditional rssi location algorithm and there is larger measurement error, real-time response is slow Shortcoming.

Brief description

Fig. 1 follows the structural representation of alignment system in real time for a kind of present invention live pick up head；

Fig. 2 is the structured flowchart of rfid reader/writer module；

Fig. 3 is the structured flowchart of zigbee network environment；

Fig. 4 is the structured flowchart of photographic head head actuator module；

Fig. 5 is the schematic diagram of the echo state network location algorithm based on rssi model.

Specific embodiment

Below in conjunction with accompanying drawing, the specific embodiment of the present invention is described in further detail:

As shown in figure 1, a kind of of the present invention follows alignment system in real time based on the live pick up head of zigbee network, including For indicating the positioning rfid label of target, the rfid reader/writer module being attached thereto, for carrying out the zigbee of wireless location Network environment and the photographic head head controlling photographic head to follow movement.Zigbee network environment includes zigbee network and positioning master Control device.As shown in Fig. 2 rfid reader/writer module includes antenna, rfid chip, microcontroller and zigbee signal transceiver module.

As shown in figure 3, what zigbee network included being made up of Zigbee network coordinator and zigbee network node Zigbee wireless network.The network environment based on zigbee of the present invention can networking as follows: this network environment by Several (>=4) are constituted based on the network sensing node 1 of zigbee chip, and its function is equivalent to the terminal in network.According to tool Body applied environment and requirement, zigbee network sensor node is uniformly positioned over indoor environment.Setting zigbee network route Node 2, its function is equivalent to the router of network.Setting Zigbee network coordinator 3, its function is equivalent to the gateway section of network Point, serves as, following in real time, the hardware communicating with positioning main controller, network coordinator is with network sensor node in alignment system The network of composition fully connected topology, the rfid collecting label position information is all sent to by all of zigbee node in real time Zigbee network coordinator.In actual applications, zigbee network routing node 2 and Zigbee network coordinator 3 can be same One node, simply different in parameter setting internally.

For stable signal transmission, the core wireless module of above-mentioned Zigbee network coordinator and zigbee network node is equal Using cc2530 chip, zigbee signal transceiver module and zigbee network environment adopt 2.4ghz radiofrequency signal.

As shown in figure 4, positioning main controller includes the communication interface modules, embedded for receiving zigbee network ambient signal The formula information processing platform and photographic head head actuating mechanism controls module.First, marked by the rfid that communication interface modules transmits Sign real-time position information；Secondly, via intelligent positioning algorithm, determine orientation regulated quantity；Finally, photographic head head actuator control Molding block adjusts the orientation angle in length and breadth of external camera head, thus adjusting photographic head focus direction, reaching to follow in real time and determining Position.For the ease of adjusting, photographic head head actuating mechanism controls module, control head to turn to by adjusting pwm mode.Positioning Main controller preferably employs double arm kernel processes units, runs android4.2 operating system, can be by wireless telecommunications, multimedia Application, remote monitoring, distributed task scheduling process etc. combine.Positioning main controller is transmitted by universal asynchronous receiving-transmitting Device (uart) connects to Zigbee network coordinator.

The above-mentioned embedded information processing platform is integrated with the echo state network location algorithm based on rssi model, according to fixed Wireless signal strength when position rfid label moves under zigbee network environment, merges, and it is fixed to be reduced by network training mode Position mean square error, thus be accurately positioned rfid label；Meanwhile, photographic head head actuating mechanism controls module is believed according to location data Breath, sends instruction dynamic adjustment head steering wheel；；

Described rssi model is:

$\mathrm{pl}\left(d\right)=\mathrm{pl}\left(d_0\right)-10\mathrm{nlg}\left(\frac{d}{d_0}\right)+\mathrm{\ϵ}---\left(1\right)$

In formula (1), d is the distance to beaconing nodes for the calculative unknown node, and n is path loss index, depends on Zigbee network environment, the signal intensity that pl (d) receives for unknown node, pl (d₀) it is reference distance d₀The signal intensity at place, d₀For the distance of reference mode to beaconing nodes, ε is position error；

Described echo state network location algorithm comprises the following steps:

1) sample phase: a number of Zigbee network coordinator or zigbee network are set in positioning region indoors Node as reference mode, after the completion of Zigbee network coordinator and zigbee network node networking, believe by the position of these nodes Breath all it is known that the mutual decay intensity rssi value of measurement, builds w_out, described w_outFor reserve pool dr(dynamic Reservior) it is the connection weight matrix inputting and exporting for output；This data is sent out by Zigbee network coordinator via uart Deliver to register control；Positioning rfid label is moved into the original state for network defined in zigbee network, that is, network is first Beginning state is 0, i.e. x (0)=0, and (u (n), n=1,2 ... are p) through input connection weight matrix w for training positioning sample_inIt is added into storage In standby pond dr, it is sequentially completed system mode and error outputCalculating and collection；Calculate output connection weight matrix w_out, Zigbee network coordinator starts to collect internal state variable from a certain moment m, and with vectorial (x₁(i),x₂(i)…,x_n(i)) (i=m, m+1 ... are p) that (p-m+1, n), corresponding sample location data ε (n) is also collected row composition matrix β simultaneously, and structure Become column vector τ (p-m+1,1)；

2) the weight computing stage: collect alignment system state matrix and positioning rfid label movement according in sample phase Sample data, calculates output connection weight matrix w_out；Due to state variable x (n) and outputBetween be linear relationship, and need Target to be realized is using echo state network reality outputApproach desired output ε (n), that is,

$\mathrm{\ϵ}\left(n\right)\≈\widehat{\mathrm{\ϵ}}\left(n\right)={\mathrm{\σ}}_{i=1}^t{w}_i^{\mathrm{out}}{x}_i\left(n\right)---\left(2\right)$

Namely wish to calculate weights(For matrix w_outElement), meet network positions mean square deviation minimum it may be assumed that

$\min \frac{1}{p-m+1}\underset{n=m}{\overset{p}{\mathrm{\σ}}}{(\mathrm{\ϵ}\left(n\right)-{\mathrm{\σ}}_{i=1}^l{w}_i^{\mathrm{out}}{x}_i\left(n\right))}^2---\left(3\right)$

Can be in the hope of the inverse matrix of matrix β according to formula (3), i.e. w_out=β^-1τ, so far, esns network training completes, training What good network can be directly used for positioning rfid label follows positioning in real time, has very high location efficiency and precision.

Claims (5)

1. a kind of alignment system is followed in real time based on the live pick up head of zigbee network, including the positioning for indicating target Rfid label, the rfid reader/writer module being attached thereto, for carrying out the zigbee network environment of wireless location and controlling shooting Head follows the photographic head head of movement, and described rfid reader/writer module includes antenna, rfid chip, microcontroller and zigbee letter Number transceiver module it is characterised in that: described zigbee network environment includes zigbee network and positioning main controller, described zigbee Network includes the zigbee wireless network being made up of Zigbee network coordinator and zigbee network node, described positioning main controller Hold including the communication interface modules for receiving zigbee network ambient signal, the embedded information processing platform and photographic head head Row mechanism controls module；

The described embedded information processing platform is integrated with the echo state network location algorithm based on rssi model, according to positioning Wireless signal strength when rfid label moves under zigbee network environment, merges, reduces positioning by network training mode Mean square error, thus be accurately positioned rfid label；Meanwhile, photographic head head actuating mechanism controls module is believed according to location data Breath, sends instruction dynamic adjustment head steering wheel；

Described rssi model is:

$pl\left(d\right)=pl\left(d_0\right)-10n\lg \left(\frac{d}{d_0}\right)+\mathrm{\ϵ}---\left(1\right)$

In formula (1), d is the distance to beaconing nodes for the calculative unknown node, and n is path loss index, depends on Zigbee network environment, the signal intensity that pl (d) receives for unknown node, pl (d₀) it is reference distance d₀The signal intensity at place, d₀For the distance of reference mode to beaconing nodes, ε is position error；

Described echo state network location algorithm comprises the following steps:

1) sample phase: a number of Zigbee network coordinator or zigbee network node are set in positioning region indoors As reference mode, after the completion of Zigbee network coordinator and zigbee network node networking, the positional information of these nodes is equal It is known that the mutual decay intensity rssi value of measurement, build w_out, described w_outFor reserve pool dr (dynamic reservior) It is the connection weight matrix inputting and exporting for output；Zigbee network coordinator sends this data to positioning control via uart Device processed；Positioning rfid label is moved into the original state for network defined in zigbee network, that is, the original state of network is 0, I.e. x (0)=0, (u (n), n=1,2 ... are p) through input connection weight matrix w for training positioning sample_inIt is added into reserve pool dr In, it is sequentially completed system mode and error outputCalculating and collection；Calculate output connection weight matrix w_out, zigbee net Network coordinator starts to collect internal state variable from a certain moment m, and with vectorial (x₁(i), x₂(i) ..., x_n(i)) (i=m, m+ 1 ..., it is p) that (p-m+1, n), corresponding sample location data ε (n) is also collected row composition matrix b simultaneously, and constitutes row Vectorial τ (p-m+1,1)；P is time serieses, and that is, same node is in not state value in the same time；

2) the weight computing stage: collect alignment system state matrix and positioning rfid label movement sample according in sample phase Data, calculates output connection weight matrix w_out；Due to state variable x (n) and outputBetween be linear relationship, and need reality Existing target is using echo state network reality outputApproach desired output ε (n), that is,

$\mathrm{\ϵ}\left(n\right)\≈\widehat{\mathrm{\ϵ}}\left(n\right)={\σ}_{i=1}^t{w}_i^{out}{x}_i\left(n\right)---\left(2\right)$

Calculate weights(For matrix w_outElement), meet network positions mean square deviation minimum it may be assumed that

$\min \frac{1}{p-m+1}\underset{n=m}{\overset{p}{\σ}}{\left(\mathrm{\ϵ}\right(n)-{\σ}_{i=1}^l{w}_i^{out}{x}_i(n\left)\right)}^2---\left(3\right)$

Can be in the hope of the inverse matrix of matrix b according to formula (3), i.e. w_out=b^-1T, so far, esns network training completes, the net training Network can be directly used for position rfid label follow positioning in real time；Esns is echo state network；T be node sample training when Between sequence；The moment that l finishes for sample training.

2. according to claim 1 alignment system is followed in real time based on the live pick up head of zigbee network, its feature exists In: the core wireless module of described Zigbee network coordinator and zigbee network node is all using cc2530 chip.

3. according to claim 1 and 2 alignment system is followed in real time based on the live pick up head of zigbee network, its feature It is: described photographic head head actuating mechanism controls module, control head to turn to by adjusting pwm mode.

4. according to claim 1 and 2 alignment system is followed in real time based on the live pick up head of zigbee network, its feature It is: described zigbee signal transceiver module and zigbee network environment adopt 2.4ghz radiofrequency signal.

5. according to claim 1 and 2 alignment system is followed in real time based on the live pick up head of zigbee network, its feature It is: using double arm kernel processes units, described communication interface modules adopts asynchronous receiving-transmitting transmitter to described positioning main controller.